(84h) A System-Level, Hybrid-AI Digital Twin Approach to Optimizing Heat Exchanger Cleaning in Separation Units with Heat Recovery

Many operations in Chemical Engineering, particularly in Oil & Gas, involve the use of thermally based separation units, such as distillation columns. These unit operations almost always include some form of heat recovery, as operational costs are largely energy-driven; thus, the more heat that can be recovered, the less energy input is required. As such, heat exchangers play a crucial role in the economic performance of these units. However, fouling is a major issue in heat exchanger operation. As heat exchangers accumulate fouling, their heat recovery efficiency decreases, leading to increased energy costs and higher carbon emissions. Excessive fouling can eventually compromise the separation unit’s performance, sometimes requiring a reduction in feed flow rate.
Heat exchangers can be cleaned to restore heat recovery capacity, but these cleanings can be costly and often necessitate shutting down the unit for several days, leading to production losses. Optimizing the cleaning schedule for heat exchangers, especially the preheaters within the separation unit, is a critical step to maximize economic efficiency. There are various methods to achieve such optimization. In this work, we present the results of applying a hybrid-AI digital twin to model both the heat exchangers and the separation unit (a distillation column) to determine an optimal cleaning schedule. This study highlights the importance of a system-level approach that considers the close coupling between heat recovery and separation units, achieving significant cost savings.