BACKGROUND
In recent decades, the chemical, refining, petrochemical, pharmaceutical, and agrochemical process industries have actively adopted distributed control systems (DCSs) to digitally monitor and control their plant operations. These commercial DCSs can provide process monitoring, control, and limited alarm management capabilities. In particular, the built-in alarm management systems require extensive expert knowledge to create an exhaustive alarm list and triggering values for each process parameter being monitored. As a result, most plants are either unequipped with any online and automated process monitoring and fault detection tools, or they have to develop in-house tools. These tools do not have the algorithmic capabilities to handle nonparametric, heterogeneous, and high-dimensional big data streams or systems with only partial observations, which are common in plant operations. Thus, there is a need for developing an advanced algorithmic framework and toolset to fill this gap in order to improve energy efficiency and safety in processing industries.
SUMMARY OF TECHNOLOGY
Researchers at OSU have developed an advanced algorithmic framework for online process monitoring and fast, accurate fault detection of distillation systems, one of the most critical unit operations in chemical processing industries. This software tool employs statistical process control techniques for multivariate process monitoring and can detect any process anomalies from industrial big data streams that are high-dimensional, nonparametric, and heterogeneous, making it ideal for industrial distillation monitoring applications. Additionally, this software tool incorporates an adaptive sampling strategy, allowing effective anomaly detection even when the system is subject to partial observations due to failing sensors, limited monitoring bandwidth, and other resource constraints. Using this allows chemical plants and refineries to embrace digitalization to save energy, reduce manufacturing costs, increase productivity, and improve process safety and sustainability in the era of Industry 4.0.
Figure 1: Phillips 66 refinery in Ponca City, OK, with red arrows showing distillation columns.
POTENTIAL AREAS OF APPLICATION
- Refineries
- Petrochemical industry
- Chemical industry
- Pharmaceutical and agrochemical industries
MAIN ADVANTAGES
- Online monitoring of big data streams
- Fast, accurate fault detection with low false alarm rate
- Works even when only partial observations are available
STAGE OF DEVELOPMENT