CEAT Research Looks At Detecting, Diagnosing Faults For Chemical Distillation Systems
Researchers in the College of Engineering, Architecture and Technology’s School of Chemical Engineering are involved in a research project to better detect and diagnose process faults in real time for chemical distillation systems.
A team led by Dr. Zheyu Jiang, assistant professor in chemical engineering, is working on a two-year Oklahoma Center for the Advancement of Science and Technology project called “FARM: Fast, Accurate, Robust Fault Detection and Diagnosis Software for Industrial Distillation Monitoring.” The team is partnered with Fractionation Research Inc., a company located in Stillwater that conducts distillation research for 75 companies worldwide.
Effective online monitoring and fast, accurate fault detection are critical to ensure safe, productive and sustainable operation of chemical manufacturing processes, including distillation, one of the most energy- and carbon-intensive processes in chemical, petrochemical, refining industries. Jiang said it is estimated that distillation contributes 40-70% of chemical plants' direct energy use and greenhouse gas emissions.
Distillation systems are monitored continuously by various sensors that provide large volumes of real-time sensor data. However, it is difficult to effectively harness that data to enable fast, accurate, and robust fault detection and diagnosis. This is because the current software cannot handle complex data structures and characteristics embedded in the online sensor measurements.
This causes chemical plants to rely heavily on human experience and heuristics to detect and raise alarms, which can cause problems to be noticed long after they have happened. Jiang’s team is developing a software program called FARM that addresses these challenges by exploring and utilizing the underlying statistical features of these complex online data streams. Programs like FARM have been used to detect solar flares early, detect malfunctions in 3D printers and detect wind turbine failures.
“By developing advanced computational techniques and algorithms such as statistical process control, FARM allows chemical plants and refineries to detect any distillation process anomaly as soon as possible,” Jiang said. “We also leverage deep learning techniques to enable accurate fault classification.”
This software could help mitigate problems in industrial distillation operations, helping a chemical plant or refinery improve its energy efficiency and save millions of dollars in operating costs annually.
“We will leverage the state-of-the-art distillation experimentation facilities and rich historical process data from our industrial partner FRI to conduct offline and online software testing and prototyping,” Jiang said.
CEAT graduate students will play a pivotal role in the project, working with FRI scientists and other industrial partners to develop, test and validate the FARM software. They will also present results at international and national conferences, engage with peer researchers and industrial practitioners and create a user-friendly interface.
“By working closely with FRI, students will also gain valuable practical experience, which will help them prepare for future careers,” Jiang said.
The team has already achieved success, as an algorithm they created helped the company’s engineers detect steady-state operations at 5-10 times faster than what has been used previously. The team has also developed a concept for detecting distillation flooding at an early stage.
“We have tested both algorithms using FRI’s actual plant data offline,” Jiang said. “The next step will be to incorporate this algorithm into our FARM software, develop the user interface and deploy the software online for real-time monitoring.
“The newly released U.S. Department of Energy Industrial Decarbonization Roadmap has identified distillation process innovation as a near-term action item to improve energy efficiency and decarbonization in chemical and refining industries. To achieve this goal, there is a huge opportunity to push the boundary of the safe operating window of distillation units toward higher energy efficiencies by implementing faster, more accurate real-time process monitoring technologies such as FARM.”
Source: Oklahoma State University