Manuel Florez on the Real World of Batch

Quest for the Perfect Batch... A problem... A classic case of pproblems encountered with automation and control without fully understanding the process...  Problem description: We had a supply crisis, and increased demand..and we were met with a product taht was failing for poor reliability, high manufacturing costs, adn porduct qulity variances. It was a multiple step process, and the last step, a batch distillation, had multiple failures-- zero output. Florez showed pictures of the flow if you aren't here, and you didn't get a copy of the CD, ppppppppth! Next year you gotta come. He showed some awful performance curves. They improved analytical methods, and increased sample frequency. They identified component 'trapping' problem, and they modified equipment and processing method. The noise in the process was so great that we decided to tackle only the most obvious first. A brilliant engineering method. Component trapping-- not typical but identified early on as a cause of product rejection...we were generating a light impurity that we could not get out of the system... so we set up a new sampling point, and we lowered the level setpoint of the reflux drum, and came up with a system flushing technique to get rid of it. Batch Analysis Approach - PAIP (process analysis and improvement pathway)
  • This flexible conceptual approach to process advancement includes methods and toosl for systematic process review, documentation, improvement, validation, and continuouus updating
  • Specifically tailored for Lyondell F&F manufacturing but easily adaptable to other manufacturing environments
  • contains elements of manufacturing to target, six sigma, process modeling and Lyondell F&F developed tools
He showed a PAIP flow chart--- and you can't see that either, nyaaaaah! Next year, show up at the conference. We set a PAIP goal for KG of good product from batch distillation. Florez showed a graph of the reduction of noise in the process.
  • Noise was found in a lot of places.
  • We had issues with data quantity, quality and odor.
  • We increased frequency and fraction cut points, with near real time material balances/live performance monitoring, and a 'sniff GC' for offensive odor identification.
  • We Worked with Equipment reliability.
  • We worked on operational issues, changing operating instructions, predicting cut sizes based on charge composition, and initial water removal.
  • We created some process performance expectations.
We made a Process Reference File-- a legacy document.
  • process description
  • process flow diagram(s) - (including storage)
  • process material balance (overall and component)
  • process parameters
  • process automation and historian
  • process performance standards (PPSs) and Charts (PPCs)
  • Operating Instructions
  • Process Sequence Tables(s)
  • Process troubleshooting
  • analytical methods and data
  • waste management
  • safety and environmental issues
Opertors were trained, on elements of the Process Reference File...they got more training they had ever seen. They got tained on batch distillation theory and troubleshooting. We initiated daily monitoring of process performance. We monitored individual batch progress. Operators entered GC sample data into electronic worksheets for data sharing. We contacted operators whenever nearing cut poitns, anyting amis was noted or to provide reinforcement of good decisions. Here's what we did:
  • We reinforced the training through discussion and application to unfolding process over months.
  • We found that the recycle management strategy was also producing noise and needed fixing.
  • There was a continuous improvement over time, except for this hidden noise.
  • We went to a more proactive recycle management strategy, returning all recycle to the subsequent crude batch.
  • We went from 30% batch failures to 0% batch failures in 2 years and 2 months.
  • There is still variability, but it can be accounted for, and now we are working back through the precursors of the process to remove the causes of the variability as we move backward.
  • We resolved the supply crisis.
  • We have now set up an approach to problem solving: PAIP
  • Operator training and monitoring is the key to achieving and sustaining gains
  • process performance improvement with negligible capital investment
  • We won the Lyondell Operational Excellence Award for 2006, and the stage is set for automated batch sequencing and more advanced control strategies.
OUr going forward goals:
  • Ultimately, a push button start/stop process
  • Our basis will be a PRF Dequence Table and process performance criteria
  • The path will be to develop an improved process model, develop a sequence logic, develop advanced control strategy (automatic feflux ration control), determine equipment requirements, complete process and capital project scopes, and implement the strategy.
  • We expect an additional 10% overall throughput improvement
  • Increased Time and Efficiency
  • Enhanced Process reproducibility