RO-TRACK: data driven predictive analytics for seawater reverse osmosis desalination plants

One of the major sources of operational cost in seawater reverse osmosis is towards membrane replacement. While membrane replacement is certainly unavoidable due to irreversible fouling, its rate can be minimized thorough operational changes. One common action to mitigate fouling is cleaning in place (CIP) -circulating chemicals on the membrane to dissolve the fouling matter. However, CIP efficacy is highly dependent on the timing and types of the foulant. Further, the current procedure to investigate the type of foulant can take months as the membrane autopsy procedure requires sophisticated analysis. Here, we present data-driven analytics that can generate insights about the potential type of foulant and provides a timeline for the next CIP based on the membrane performance parameters. This approach is based on our insight's generator framework developed using historical data, coupled with machine learning algorithms and operational knowledge tree.