CC-95-127 Multiple Boiler Emissions Optimization Using Neural Networks

Description:

This paper presents an application of neural network technology at Citgo’s Corpus Christi Refinery. Individual neural network emissions models have been implemented on-line for each of eight utilities boilers. By considering the eight boilers together as a single emissions source, environmental compliance can be achieved while optimizing operation. A multi-boiler optimizer has been implemented on-line which uses the predictions from the individual boiler models and adjusts boiler loads to minimize refinery utilities operating costs while meeting demand and achieving environmental compliance.