Abstract: The structure of food webs is complex, but it can be analyzed through modules. Body size is known to be a good predictor of trophic position. Many studies have shown that a size structure exists within food web modules. But, the role played by the surrounding medium has not really been explored in the context of predator-prey interactions. Hence, The physical components of the surrounding medium lead to mechanical constraints on species persistence and interactions. Since predation usually implies motion, these factors create mechanical constraints acting on predators. These constraints are size-dependent. The present study investigates how physical factors from the medium can constrain the size structure of several food web modules. Hence, we built a model in which species motion and species interactions are constrained by physical properties of the medium and biological traits (e.g., metabolism). As key physical factors of the medium, we consider gravity, medium density, body density, and medium viscosity. These factors, in relation with body size, constrain the occurrence of trophic links. We consider several classical food web modules: one predator – one prey, one predator – two prey, two predators – one prey, and a three-level food chain with possible omnivory. We analyze which predator and prey sizes allow each module to persist. Preliminary results show that the dynamics of the system are mostly driven by predator size. Systems with small predators – prey size ratios lead to a single equilibrium point, while system with large size ratios show persisting oscillations. Larger predators usually outcompete smaller ones in two predators – one prey systems, while smaller prey usually outcompete larger prey in one predator – two prey systems. For food chains, the size ratio between intermediate consumers and basal prey is on average greater than the ratio between top predators and intermediate consumers. Lastly, omnivorous top predators can persist only within a narrow range of sizes. Model predictions fit data remarkably well, for a large number of species, in aquatic systems. The real novelty of this approach is that it merges size-related biological and physical constraints within classical predator-prey modules. Therefore, the size structure of the modules emerges from traits measured at the individual level. Most parameters in the model are related to predator and prey sizes, a trait that is commonly measured. This study provides new insights about size structure of food web modules.