Operational Optimization of Processes with Multistream Heat Exchangers Using Data-Driven Predictive Modeling

Abstract

Multistream heat exchangers (MHEXs) facilitate simultaneous heat exchange between multiple streams and are mainly used in energy-intensive cryogenic processes. Reducing the energy consumption of existing processes with MHEXs is important, but system-wide operational optimization necessitates that the heat-transfer parameters of the MHEXs are known. However, most MHEXs are practically black-boxes due to their proprietary designs and complex geometry. In this work, we present a procedure for the operational optimization of processes with MHEXs. Our procedure involves the development of a predictive model for MHEXs as the first step, followed by the illustration of operational optimization. We begin with the development of a data-driven nonlinear programming (NLP) model to synthesize an equivalent network of simple two-stream heat exchangers that best represents the operation of an MHEX. We then demonstrate our predictive modeling procedure on the main cryogenic heat exchanger (MCHE) from an existing natural...