Synthesis and Design of Resilient HENs

Research on the synthesis of economically optimal HENs has been performed by various investigators for over 15 years (Nishida et al., 1981). Several powerful synthesis methods have evolved, including the pinch design method (Linnhoff et al., 1982 Linnhoff and Hindmarsh, 1983) and methods based on structural optimization [to predict a minimum set of stream matches (Papoulias and Grossmann, 1983) and to determine the most economical network structure (Floudas et al., 1986) from the predicted matches]. However, these methods synthesize networks only for fixed, assumed nominal values of any uncertain supply temperatures and flow rates and uncertain heat transfer coefficients. 

The analysis techniques presented earlier in this chapter can be used to test the resilience of a synthesized network, and evolutionary changes can be made to the network, if necessary, to improve its resilience. However, this type of procedure may require many evolutionary synthesis-analysis iterations, and it may not be obvious which evolutionary changes are required to improve a network s resilience. Obviously, better methods are needed which incorporate resilience into the synthesis procedure itself. 

For simple problems with a small number of streams and a small number of uncertain parameters, the class 1 FI target can be determined simply by trial-and-error plotting of the composite curves (Colberg et al., 1988). The size of a class 1 uncertainty range is generally limited by its case B 

Trial-and-error plotting of composite curves is impractical for problems with a large number of streams, a large number of uncertain parameters, or correlated uncertainties. Colberg et ai, (1988) present a nonlinear program (NLP) to calculate the class 1 FI target for these (and simpler) problems. 

Example 13 (from Colberg et al., 1988). Consider the nominal stream data given in Table IX. Stream Sc2 causes the pinch for these data. 

---