ABSTRACT
In Chapter 10, the ProcIndustries digital transformation team extracts and exposes highly granular time-series data for corporate big data analytics, models, and machine-learning (ML) algorithms to model and predict refinery/plant process and equipment behavior. Using the EIDI’s online analytics to contextualize data makes the data more effective when using offline analytics and ML. This is because data streams become linked to physical assets, allowing workers to compare and analyze similar event-framed data. Time-series operations data are combined with other types of data for real-time situational awareness and powerful offline analytics. Because standard extraction methods are developed to prepare and transmit contextualized refinery data to the analysis software systems, time spent extracting, preparing, and formatting EIDI data is significantly reduced. Published data sets are produced correctly and can be ingested by the analytic of choice, reducing time to insight. By using an offline digital twin, engineers can develop analytics or run modeling software that simulates the process and makes predictions. Plant operations personnel can subsequently observe the variance between the predicted values and actual plant performance.
