This paper presents a process synthesizer capable of determining the optimal network configuration of complex notation plants including regrinding units and upgrading one mineral value. The set of possible network configurations is derived from a general superstructure that features all the potential connections between notation and regrinding units. The formulation of the optimization problem results in a mixed integer non-linear programming (MINLP) problem. The objective function driving the network synthesizer is the maximization of the plant annual margin of revenues-costs. Capital and operating costs relative to the processing units and their connections are accounted for. The original optimization problem with non-linear constraints is decomposed into sub-problems and reduced to the solution of programmes with a simpler set of constraints that are solved in sequence. Empirical design methods and practitioners＇ rules are implemented to generate feasible starting solutions and to post-process the connection matrices for improving the value of the objective function. The success of the procedure depends on the robustness of the notation and regrinding unit models and their mineralogical-specific rate constants. Examples of an applications demonstrate the optimisation procedure in detail.