Technical debt is a concept applied to decisions taken to favor rapid production, at the expense of future capacities to evolve or improve the product or service. This concept is transversal to all sciences, especially in experimental physics and systems. However, technical debt is difficult to manage because there is a lack of time, expertise, or knowledge on how to perform it. This paper proposes a computational model of technical debt management in software architecture, based on observations made in software teams in their daily work. This computational model exploits natural language processing and model-testing techniques on software project artifacts to build a model that allows localizing and visualizing the impact produced by the technical debt and its payment strategies. This proposal aims to support teams of architects to explain the current and future impact of the debt injected as a result of decisions made.