Intelligent Manufacturing € August € 1996 € Vol. 2 € No. 8

Eli Lilly Improves Yields with Intelligent System

Eli Lilly's Tippecanoe Laboratories (Lafayette, Ind.), a fermentation manufacturing facility specializing in bulk antibiotics, has implemented intelligent system software to achieve significant cost savings in its fermentation operations. Within one month after deployment, process deviations had decreased by 72.3% and yields had increased by 3.2%.

Tippe Labs chose G2 Diagnostic Assistant (GDA), a real-time expert system application from Gensym Corp. (Cambridge, Mass.) to help achieve three goals: control its fermentation process by reducing variability, increase yield while reducing unit costs, and transfer ownership of the process from scientists to operators.

Using statistical information from previous fermentation batches, Tippe Labs developed control charts in their data historian. The GDA application monitors incoming production data, compares them against the established control charts, and provides operators with expert advice. "If the new data point is outside the control limit, the intelligent system warns an operator about a potential deviation," explained Victor Huynh, a process engineer at Tippe Labs. "It also tells the operator how much time they have to correct the condition before it becomes an actual deviation."

When alarms occur, the intelligent system looks at the current process condition and uses its knowledge base to suggest the optimum troubleshooting technique. This provides Tippe Labs with a consistent way to address problems, based on proven expert knowledge.

The GDA application runs on an HP 9000/725 workstation communicating with a historian residing on an HP 9000/827. The HP 1000A equipment that serves as a distributed control system sends fermentation process data to the historian, allowing GDA to process data and provide feedback and expert advice in real time.

In the first month of system operation, process deviations dropped 72.3% and yields increased 3.2%. As a result, the company has realized significant cost savings. "One of the major benefits we've realized is the ability to meet increased customer demand without any additional investment in equipment or facility expansion," Huynh said.

The use of intelligent software also allowed the lab to transfer ownership of the process from scientists to operators. "With our knowledge-based application, operators receive expert assistance, and we've been able to free up our scientists for more value-added tasks," Huynh said. "Process knowledge was previously confined to an individual. If that individual left the position, the knowledge was lost, and a new operator had to learn the skill from scratch."

Tippe Labs plans to further expand the application, utilizing such intelligent technologies as fuzzy logic and neural networks to achieve an even higher level of analysis and modeling capabilities.

Click here to return to Table of Contents for the Intelligent Manufacturing August issue.