Engineering analysis has seen major advances with the development of the finite element method and its implementation in a computer form. in this study, the application of expert systems to develop computer-based assistants in conducting finite element analyses is investigated. this is based on the observation of experts through knowledge acquisition involving knowledge elicitation, in order to define the characteristics of the knowledge-based problem solving. after a review of the different knowledge elicitation methods, a particular one has been selected. it involves unstructured and talk-through interviews. ît is followed by the generation of transcripts, a concepts table and concepts graphs, which describe the knowledge to solve problems in the field. an analysis of the knowledge elicitation material is then described, where the characteristics of the experts" approach are established. the study defines the expert behavior as being based on: a meta-levei sophisticated control mechanism, tasks and model transformation. the met level control mechanism is capable of performing context definition, understanding, and task selection, and can be linear for routine analyses and complex for original cases. tasks represent the specialization of the process, and are constituted of many knowledgebase modeling operators manipulating many different knowledge types. they transform the model, which is a representation of the problem state and its evolution. the general characteristics of the knowledge-based environment are established, and the development of an expert system supporting the separation between the model, the knowledge-base and the control mechanism is proposed. the study also discusses knowledge representation issues, and defines a scheme based on procedural calls, rules and frames. the study also describes the implementation of an expert system for routine analyses. amongst its most important aspects are tasks, the model representation and a rigid control mechanism.