Abstract (EN)

Blending problems aim to select a subset of inputs from a set of existing ones and to define the input quantities to be mixed in order to produce a particular output using a predefined set of outputs. The structure of these existing inputs, defined by the % of some components in their total weight, is known. The structure of an output varies according to the chosen mixture which is constrained by % ranges that must be respected for each component. Reverse blending is an extension of the blending problem in which the set of inputs does not preexist but must be defined to be able to produce any output. In this problem, never addressed to our knowledge, the number of inputs to be produced must be as small as possible and the composition of each input is to be defined. This problem arises in the context of a growing diversity of fertilizers to be managed both in production and in distribution logistics. A real-life application example of this new approach is provided.