||Common approaches to scale-down freeze-thaw systems are based on matching time-temperature profiles at corresponding points; however, little is known about the differences in anisotropy between the 2 scales. In this work, computational fluid dynamics modeling was used to investigate these differences. The modeling of the convective flow of the liquid phase within ice porous structure and volume expansion caused by freezing enabled accurate prediction of the local temperature and composition, for evaluation of potential stresses on protein stability, such as cryoconcentration and time in the nonideal environment. Overall, the small height of the scale-down containers enhances cryoconcentration. The time under stress was consistent in both scales, except when the walls of the container could deform. In general, the common approach of matching the time-temperature profile at the center of the containers was more effective as a worst-case scenario than a scale-down model. This work shows that instead of considering a single matching time-temperature location; one should aim for a more general perspective by measuring many locations. Container geometries and heat transfer rates should be designed to match stresses related to protein integrity for equivalent mass fractions between both scales, which can be achieved with the assistance of computational fluid dynamics models. (C) 2020 American Pharmacists Association (R). Published by Elsevier Inc. All rights reserved.