Coarse-Grained Molecular Dynamics Simulations for Predicting Rheological Behavior of Casein Micelle Dispersions

Handling dispersion of casein powders in water is widely encountered in the milk industry. However, in silico prediction of the apparent viscosity of these colloidal dispersions is not an easy task, especially when these micellar casein suspensions are highly concentrated, as in hyper-protein milk beverages, which are experiencing exponential market growth. In this work, Coarse-Grained (CG) models using Lennard-Jones potentials to model interactions were built for simulating rheological properties of colloidal micellar casein dispersions (native and demineralized). In a first approach, a polydisperse explicit CG model was developed. For this polydisperse CG model, the representation chain was composed of four large smooth spheres of different sizes mimicking the real distribution of casein colloids. The CG simulation results were validated by comparison with experimental rheological data for native colloidal casein dispersions. Both in-house experimental results and available data found in the literature were used for this purpose, covering a wide range of casein concentrations ([10 g/L–200 g/L], [8–20%] corresponding to casein concentration, colloid volume fraction and solid/liquid volume fraction, respectively). In a second approach, a simplified model using a monodisperse CG model was developed. This simplified model only included one type of soft sphere and was found to preserve the accuracy of the rheological prediction. Finally, a monodisperse CG model was set up to predict the behavior of demineralized micellar casein dispersions, for which a decrease in the average size of the micelle size distribution is observed when demineralization occurs. For all models, the comparison between the predicted and experimental rheological behavior is fully satisfactory, proving that the CG models proposed for casein-based micellar dispersions are physically well founded and that the proposed simplified representation chain, based on micelle size observation, makes sense.