The general distillation sequence synthesis problem featuring the separation of multicomponent feed streams into multicomponent products is addressed. Potential flowsheets include stream bypassing and mixing and use sharp separations as well as non-sharp splits where key component distribution is allowed. Compared to conventional sharp distillation sequence synthesis, this leads to a mixed-integer non-linear programming problem of increased complexity, including non-convexities as well as multi-modalities. Product specifications create additional constraints while simultaneously call for a rigorous modeling of the non-key distribution. A synthesis method is proposed that models the various flowsheet configurations with a new and flexible superstructure concept and connects the gradient-free optimization technique of application-orientedly developed Evolutionary Algorithms (EAs) to the rigorous modeling capabilities of the Aspen plus (TM) simulation system, thus enabling realistic process design and cost objective function calculation. The re-examination of two published examples illustrates the applicability and the potential of the approach. (C) 2007 Elsevier Ltd. All rights reserved.