(188k) Optimal Operation of Heat Exchanger Networks through Heat Duty Redistribution Using Energy Flow Graphs

Energy integration is a key strategy employed to improve energy efficiency of a chemical complex and heat exchanger network (HEN) is a manifestation of such integration. Over the last 4 decades, significant amount of research effort has been dedicated for the design of HENs to achieve optimal structure (e.g. maximum energy recovery, minimum total cost, etc.) [1]. During the design stage, the hot and cold process streams are matched such that most of the energy required to heat a cold stream is provided by a hot stream using process-to-process heat exchanger (PPX) and the remaining small amount of energy is provided/rejected by an auxiliary heater/cooler. This designed operating point is then considered a target and a regulatory control system involving PID controllers is formulated to realize these targets during operation by using the duties of auxiliary heaters and coolers (and if available, any intermediate bypass streams). Operation of these HENs is subject to feed disturbances, target temperature changes as well as gradual performance deterioration due to scaling. As a response, the regulatory control layer rejects these disturbances by manipulating the auxiliary heater/cooler duties, thereby increasing the overall energy consumption of the network from the designed optimal value. These PID controllers have a local view of the impact of disturbance and in many cases, can lead to manipulated input saturation. A rational approach is to re-optimize the HEN (preserving the structure) online and use PID controllers to drive the HEN to the new optimal operating point. Small computation time and scalability to large networks are key challenges in the formulation and subsequent implementation of these optimal operation schemes.

In this talk, a hierarchical framework for optimal operation of HENs is presented. At the heart of the framework is a graphical representation of a HEN in which each of the exchangers (PPX as well as auxiliary heaters/coolers) are considered as nodes and energy flows connecting them constitute the edges. Similar representation has been previously used to reduce a maximum recovery network [2]. Elementary cycles (closed loops) are identified in these graphs and each loop is given an additional flexible load which allows for redistribution of heat duty across exchangers. One of the loop loads is related to the external utility and the optimization problem is set up for its minimization. The resulting problem is a constrained LP optimization which can be solved within few seconds. The output of the optimization (loop load values) are then converted into intermediate temperature setpoints and downloaded to the lower regulatory PID control layer. The graphical representation can incorporate feed disturbances as well as performance deterioration across a set of exchangers. The graphical representation allows scalability to large networks as the size of the optimization problem scales linearly with the number of exchangers. The effectiveness of the proposed scheme is demonstrated through closed loop simulations of a 4 stream, 8 exchanger HEN under various operating scenarios.

References:

[1] Furman, K.C. and Sahinidis, N.V. A critical review and annotated bibliography for heat exchanger network synthesis in the 20th century. Ind. Eng. Chem. Res., 41(10), 2335-2370, 2002.

[2] Mehta, R.K. and Devalkar, S.K. and Narasimhan, S. An optimization approach for evolutionary synthesis of heat exchanger networks. Trans. IChemE, 77A, 143-150.