Purpose: This article deals with modelling and design of embedded autonomous microsystems able to harvest energy from their close environment. Due to their numerous functionalities, progress in electronics and the development of wireless applications, microsystems are used in a wide range of applications. Moreover, they require a large autonomy to ensure reliability and avoid maintenance operations. We focus on energy harvesting to power them by replacing the conventional energy source with an energy harvester. Different sources can be used but one difficulty is to select the most adapted to a specific application and choose the energy harvester architecture to get an efficient system. Method: A decision support to design autonomous microsystems working by harvesting energy is proposed. It aims to support designer's decisions from qualitative representation to physical models. Our multidisciplinary approach is based on the identification, analysis, modelling and minimization of antagonist flows and effects through a system-level model to estimate the energy that can be harvested. Result: The method is illustrated through a validation case dealing with the displacement and deformation measurement of a cantilever beam with an accelerometer. The aim is to choose the energy source and the architecture of the energy harvester to satisfy the requirements. Discussion & Conclusion: Through the application, this paper demonstrates the relevance of our tool in aiding the design of the energetic part of an autonomous embedded microsystem. We can also note that our top-down approach can be use as a decision tool.< Réduire