A knowledge-based system for the supervision of a wastewater treatment plant was successfully applied to a full-scale facility. The key factor of this supporting tool development was the two-phase methodology used to acquire and fix the knowledge into the knowledge base. Both phases of the methodology are presented in the paper; the first consists of literature reviews and Bite interviews with domain experts, while the second is based on machine learning tools and is subdivided into four steps: data handling, classification, interpretation, and codification. The aim of this two-phase methodology is meant to ease the knowledge acquisition process. The main objective is to find the relevant issues and then reduce the space of search according to the target facility. Also, this methodology allowed the user to explore the data space to discover, if any exist, new pieces of knowledge. This methodology can be generalized to acquire specific knowledge from any (bio)chemical process, improving the development process and the efficiency of the supervisory knowledge-based system.