An algorithmic approach was used to select recoveries for non-sharp sequence designs. Simple distillation columns were employed. The non-key component distribution was modeled using the Fenske equation. The combination of producing impure products and allowing non-key component distribution, results in a large search space. With this large search space size and software limitations, five problems were solved (see summary table in the Design Examples sub-section). The optimal non-key component distribution was found to be significant but not necessarily maximal. Parallel processing was selected as optimal in contrast to the more traditional sequential processing.