A best-first search strategy for energy relaxation in MER heat exchanger networks

The pinch design method efficiently generates a maximum energy recovery (MER) network which meets the utility targets for a given value of the the minimum approach temperature difference. However, this MER design usually contains a significantly greater number of heat exchanger units than the theoretical minimum. Loop breaking and energy relaxation may be used to eliminate these additional units. At each stage of loop breaking, it has been recommended that the unit with the smallest heat load should be removed. However, this study shows that this heuristic can lead to suboptimal designs with respect to energy consumption. An alternative systematic method is presented which reduces the number of units such that the energy penalty is a minimum. A mixed integer non-linear program (MINLP) is formulated with the objective of minimizing the energy consumption for a given number of units. Subsequent loop-network interaction analysis helps in identifying the exchanger units which are good candidates for removal. A lower bound on the consequent energy penalty is also evaluated. These bounds are employed in a ‘best-first’ search strategy to solve the proposed model. On removal of the candidate unit/units, the resulting topology transforms the MINLP to a non-linear/linear program (NLP/LP) which is solved by conventional algorithms. A loop and path identification algorithm (LAPIT), based on graph theory, has been developed as an aid to these computations.