In fault diagnostic expert systems, the knowledge can be either shallow (experience-based) or deep (function-based). This paper presents a deep-knowledge expert system for fault detection in process operations and control domain. The structure used for modelling deep-knowledge is called Goal-Tree-Success Tree. An expert system shell has been built for applying this deep- knowledge model utilizing the logic based language PROLOG. An example, applying this deep-knowledge model and the developed expert system shell, is also presented.