Work and Heat Exchange Networks – Opportunities and Challenges
Journal article, Peer reviewed
MetadataShow full item record
Original versionComputer-aided chemical engineering. 2018, 44 481-486. 10.1016/B978-0-444-64241-7.50075-6
Design and optimization of Heat Exchanger Networks (HENs) is well established and has been heavily applied in the process industries since its pioneering period in the 1970s and 1980s. While temperature and thermal energy (heat) obviously are key elements in process plants and power stations, pressure and mechanical energy (work) are equally important. By adding expanders and compressors as equipment and power as a utility to the classical HEN problem, the considerably more challenging Work and Heat Exchange Network (WHEN) problem has been formulated as a new and growing discipline within Process Systems Engineering (PSE). Considerable opportunities exist for improving energy efficiency in process plants and power stations by using the WHEN methodology. Recent applications include (I) design of LNG processes, (II) design of CO2 capture processes, and (III) an industrial sensible heat pump. In fact, WHENs have strong similarities to Heat Pumps and Refrigeration Cycles, with the additional advantage of a PSE approach. Challenges in the graphical methodology for WHENs compared to HENs include (a) the shape of Composite and Grand Composite Curves will change with pressure manipulations, (b) the Pinch points may change, (c) the hot and cold utility demands will change, (d) the stream identity (hot or cold) may temporarily change, (e) process streams may be used as utilities, and (f) work and heat have different energy quality. To overcome the above challenges, the WHENs problem could be addressed by developing Optimization models using Total Annual Cost as the objective function. However, new challenges will appear if Mathematical Programming is used, such as (i) developing a sufficiently rich yet efficient superstructure, (ii) nonconvexities in the model that may result in local optima, (iii) potentially a large number of binary variables that may give a combinatorial explosion, and (iv) discontinuities in the process models that will cause numerical problems and slow down convergence, unless recent developments in nonsmooth analysis can be adopted for these problems. © 2018 Elsevier B.V.