(178a) A Chemical Industry Case Study: Using AI to Create More Productive Teams in the Chemical Manufacturing Industry

The development of an effective and efficient search functionality based on AI looked to solve two challenges cited by Bayer:

1. Inefficient access to documented historical knowledge captured by Shiftconnector; and
2. Incomplete entry items where often descriptions of solutions to problems (in the production process) were missing or incompletely described. The development of AI’s Smart Search functionality solved both problems. Teams have improved plant performance by quickly finding solutions based on the history of documented tribal knowledge from plant operations teams. By retrieving the most relevant information, it accelerates operations and brings capabilities to help teams recover from disruptions by immediately identifying appropriate fixes. Also, captured knowledge can be managed to support workforce changes.

Eschbach's questionnaires and analysis of usage data show the following advantages of Smart Search:

• Smart Search is well accepted and used several times daily.
• The time users spend searching for information decreased significantly. Instead of searching for several minutes, users get results within seconds.
• User statistics show that users access all historical knowledge (> 8 years) captured in Shiftconnector. Prior to the Smart Search introduction, only records from the last month were accessed.
• Statistics and questionnaires show that now the historical know-how captured in Shiftconnector (solutions to problems in the production process) is effectively used by all users of Shiftconnector, from shop-floor operators to process engineers.

The Shiftconnector Smart Search feature changes the way shop floor operators in the process industry search. The Shiftconnector Smart Search feature was developed using the latest technology and architecture for AI-based language models. This allows the Smart Search feature to understand the context of the search versus a standard keyword search that would need very specific terms to retrieve relevant information. With Machine Learning, the smart search retrieves relevant information with a simple, everyday phrase like “why is this product brown instead of gray?”

Search engines in today’s market like Google are smart but do not think like someone in the process industry as they lack industry-specific or even plant-specific terminology.

Shiftconnector’s Smart Search is superior in this aspect as it was developed with a lexicon of the operators in the process industry. With nearly 20 years in the industry and 70,000 global users, the eschbach team understands not only the specific use cases for a search engine but also the specific terminology that is used every day on the plant floor. Consequently, Smart Search covers plant-specific terminology, such as searching for technical labels which are only used at a certain production line. For example, if in a plant the label “CP001” is used for a centrifugal pump at a reactor inlet, Smart Search understands this and will retrieve relevant information linked to “CP001” when an operator is searching for “centrifugal inlet pump”.