Saccharomyces cerevisiae is a broadly used yeast form in experimental biology. Thanks to its relatively short lifecycle and its robustness in a variety of environments, it is a particularly good candidate for experimental evolution. This field focuses on the population dynamic of yeast in different growth media. However, the biological tools currently at hands do not always provide satisfying insights on the observed dynamic during those experiments. As a result, the development of numerical twin of yeast populations through the calibration of suitable mathematical models can greatly improve our understanding of the observations. We are interested in a mass-structured growth-fragmentation model for the population dynamic of a unicellular strain of Saccharomyces cerevisiae. Integral properties of such a population can directly be derived from the mass distribution and may help calibrating the model onto experimental data. Nonetheless, the numerical resolution of such models may often encounter consistency issues with respect to one or several of its integral properties. In this talk, we will introduce a ENO-based finite volume scheme corrected with weights, which allows us to simultaneously retrieve the model’s solution and its two first moments. The method will then be used to calibrate several growth-fragmentation models on experimental measures, thus characterizing a yeast cell population in fermentation phase. This study opens the way toward consistent numerical twins of populations ruled by growth-fragmentation.