Architectural Technical Debt (ATD) is a metaphor used to describe consciously decisions taken by software architects to accomplish short-term goals but possibly negatively affecting the long-term health of the system. However, difficulties arise when repayment strategies are defined because software architects need to be aware of the consequences of these strategies over others decisions in the software architecture. This article proposes REBEL, a semi-automated model-driven approach that exploits natural language processing, machine learning and model checking techniques on heterogeneous project artifacts to build a model that allows to locate and visualize the impact produced by the consciously injected ATD and its repayment strategy on the other architectural decisions. The technique is illustrated with a data analytics project in Colombia where software architects are unaware of the consequences of the repayment strategies. This proposal seeks to support teams of architects to make explicit the current and future impact of the ATD injected as a result of decisions taken, focusing on the architectural level rather than code level.