Abstract

A virtual environment (VE) of an industrial splitter has been developed to test a dynamic real-time optimizer prior to on-site deployment. The paper describes the architecture of the VE, which represents the process using the EcosimPro (c) modeling and simulation software, and a real-time manager to maintain a controlled pace of simulation progress. The virtual process has been integrated with the same commercial predictive controller (Aspen (c) DMC) as installed in the plant (even with the same configuration) and a dynamic real-time optimizer has been developed in Matlab (c) and implemented on top of it, allowing the VE to test the optimization layer before deployment in the real factory. The data generated by the plant is stored using InfluxDB (c), a time series data platform, for further analysis. The results show that the VE allows revamp studies to be conducted with less cost and more confidence.