In software engineering, the separation of concerns is an important measure for the maintainability of a project. If multiple domains are mixed in an artifact, all of them must be understood before changes can be made. This problem is aggravated for Research Software Engineering (RSE), since a high portion of researchers lack software engineering training, but increasingly write complex research software. Thus, they might be more prone to mix several concerns in one artifact, such as scientific code and technologic code. Additionally, they might have a harder time understanding technical aspects of a project, such as the database integration. A partition of a research software projects into technical and scientific code is therefore essential to help researchers focus on their domain. For classical software engineering the basic concept of separating code based on purpose and a accompanying methodology was already demonstrated with Quasar [4]. The project Wissenschaftsinduzierte Retrospektive Code-Analyse(WiRCA), a RWTH-internal cooperation between the Chair of Science Theory and the Chair of Software Engineering, aims to apply this separation to research software. It focuses on identifying mixed purpose, hard to understand artifacts, which should be refactored to improve the project structure and thus maintainability. The resulting methodology and tooling has been implemented in a demonstrator and was applied to GemPy [2], a research software project for the 3D modeling and analysis of geological structures.