Multiphysics version 5

For the numerical computation of the above-listed models, the finite-element method (FEM)-based simulation software COMSOL Multiphysics^® version 5.6 (COMSOL Multiphysics GmbH, Zurich, Switzerland) was used.
As explained in Section 2.2.1, the dielectric parameters for the SC layer are interpolated with Equation (16). In [22 (link)], the epidermis/dermis layer was approximated by blood and the hypodermis by infiltrated fat. The dielectric parameters for these two layers were taken from [7 (link)], as listed in Table 2.
For the evaluation of the water concentration in the SC and the reproduction of the measurements in [21 (link)], the dynamics of water transport in the SC model was used (1)–(6). The SC was initialized as fully hydrated and exposed to a relative humidity of 30% until equilibrium was reached. The 30% of relative humidity was chosen to simulate a realistic dry environment. The reached equilibrium then served as initial SC hydration for the simulation of the hydration boundaries and other hydration levels, as investigated in Section 3.1. Table 3 summarizes the initial conditions and boundary conditions.

Computational fluid dynamics simulations (CFD) were performed to analyze nutrient supply inside the cultivation chamber designs as described before^{13 (link),27 (link)}. For the CFD simulations COMSOL Multiphysics Version 5.6 (COMSOL AB, Sweden) was applied. Glucose concentration profiles during MSCC, assuming a steady state with 181 cells inside the chamber with a constant glucose uptake rate of 3800 nmol per 10⁶ cells and day, were simulated under consideration of the respective cultivation chamber geometry. Cellular uptake was assumed to be homogeneously distributed over the whole cultivation chamber. Glucose concentration inside the supply channels was set to 45 mol m⁻³.

For the estimation of fluid flow pattern and its mechanical effects, in silico
modelling was undertaken using COMSOL Multiphysics version 5.6 (COMSOL AB,
Sweden). Briefly, the geometry of the culture chamber including the scaffolds
and the rectifiers was reproduced computationally. The mechanical properties of
the culture medium were assumed to be comparable with water. The scaffolds were
assigned as porous domains so that the Darcy’s law was applied to avoid
excessive computational burden. Porous properties were parameterised using data
previously obtained by microCT analysis of the scaffold, including porosity
91.708% and permeability 5.80216e-09 m².^{29 (link)}
From the inlet to the outlet, a fully developed flow was applied at a
flow rate of either 0.8 ml/min (FL-L) or 1.6 ml/min (FL-H). Non-slip wall
condition was applied to the boundary condition of the fluid paths (i.e. metal
parts, silicon tubes). As a representative for visualisation of shear stress
within the complex 3D structure, the geometry of the scaffold was obtained by
microCT (Skyscan 1172, SkyScan, Belgium) primarily and converted into an
.stl file. A cube with a diameter of 1.2 mm was dissected
from the middle of the geometry to make it possible to proceed with computation.
The dissected part was assumed to be located at the front row of the scaffold
pile.