We design a stochastic individual-based model structured in energy, for single species consuming an external resource, where populations are characterized by a typical energy at birth in $\mathbb{R}^{*}_{+}$. The resource is maintained at a fixed amount, so we benefit from a branching property at the population level. Thus, we focus on individual trajectories, constructed as Piecewise Deterministic Markov Processes, with random jumps modelling births and deaths in the population; and a continuous and deterministic evolution of energy between jumps. We are mainly interested in the case where metabolic (i.e. energy loss for maintenance), growth, birth and death rates depend on the individual energy over time, and follow allometric scalings (i.e. power laws). Our goal is to determine in a bottom-up approach what are the possible allometric coefficients (i.e. exponents of these power laws) under elementary -- and ecologically relevant -- constraints, for our model to be valid for the whole spectrum of possible body sizes. We show in particular that assuming an allometric coefficient $α$ related to metabolism strongly constrains the range of possible values for the allometric coefficients $β$, $δ$, $γ$, respectively related to birth, death and growth rates. We further identify and discuss the precise and minimal ecological mechanisms that are involved in these strong constraints on allometric scalings.