AI-Supported Workflows for Chemical Batch Plants

This article presents an AI supported solution for cross-application workflows in batch plants. Based on a digital virtual assistant, industrial digital services are connected in workflows and support users in making the best use of the digital infrastructure. We focus on two digital services which allow for early detection of faults within the production and root cause analysis: a novel approach to online identification of batch failures building on an adjusted form of multiway principal component analysis and a low-cost sensing approach to perform root cause analysis for different failure modes occurring in batch plants. A case study of the implementation in a test plant is reported together with insights into the benefits and limitations of the approach gained via several process executions.