Computer-aided molecular design has emerged as a powerful technique to identify promising compounds that meet the predefined property targets. These techniques have been employed in various areas, including the design of solvents, refrigerants, and polymers.

This paper will present a mixed-integer non-linear program (MINLP) for designing a secondary refrigerant. Secondary refrigeration loops have recently been recognized as a potential solution to reduce the environmental impact of retail food refrigeration. The MINLP model adopted here includes a property prediction model (Nanda, 2001) coupled with molecular structure feasibility constraints (Sahinidis et al., 2003). A large number of basic building groups is considered and no limits on the type of the molecule are imposed. To cope with the numerical challenges associated with nonlinearities of this model, a direct enumeration of the search space is used to solve the model for restricted molecular sizes, while a branch-and-bound algorithm is used for larger designs. Several novel solutions are obtained and will be presented and analyzed in detail. We also explore the possibility of using accurate property models (Marrero and Gani, 2001) along with the possibility of extending these models to include higher-level properties which are structure-dependent for larger molecules.

Bibliography:

Marrero J. and R. Gani, A group contribution based estimation of pure component properties, Fluid Phase Equilibria, 183, 183-208, 2001.

Nanda, G., Design of efficient secondary refrigerants, M.S. thesis, University of Illinois at Urbana-Champaign, 2001.

Sahinidis, N. V., M. Tawarmalani, and M. Yu, Design of alternative refrigerants via global optimization, AIChE Journal, 49(7), 1761-1775, 2003.