We study the anomalous diffusion version of the quasistationary Stefan problem (the fractional quasistationary Stefan problem) in the multidimensional case Ω(t) ⊂ R n , n ≥ 2. This free boundary problem is a mathematical model of a solute drug released from a polymer matrix (n = 1, 3). We prove the existence and uniqueness of the classical solution for this moving boundary problem locally in time. A numerical solution is constructed in the two-dimensional case.
MathematicsSubject Classification. Primary 35R35, 35C15; Secondary 35B65, 35R11. Keywords. Quasistationary Stefan problem, Anomalous diffusion, Caputo derivative, Coercive estimates. 544 N. Vasylyeva and L. Vynnytska NoDEA tional velocity of the boundary Γ(t) in the direction of the normal n t and is represented by (see, e.g., [44]):(1.5) D ν t denotes the Caputo fractional derivative with respect to t and is defined by (see (2.4.6) where Γ(ν) is the Gamma function, < ·, · > denotes the scalar product. We recall that the anomalous diffusion can not be modelled as a standard Brownian motion [8,35], and the mean square displacement of the diffusing species (Δy) 2 scales as a nonlinear power law in time, i.e. (Δy) 2 ∼ t ν for some real number ν. In the case ν ∈ (0, 1), this is referred to as a subdiffusion.If n = 1, 3 and Γ 0 is in the exterior of Γ(t), problem (1.1)-(1.4) describes a solute released from an undissolved monolithic polymeric matrix in the quasistationary case [28,29]. We recall that each matrix consists of two regions: the surface zone (or liquid phase) x ∈ Ω(t) and the core B(t) within the interior of Γ(t), and B(t) ∩ Ω(t) = ∅. The whole solute is dissolved in the surface zone, and the core contains undissolved solute. The two zones are separated by the diffusion front (liquid-solid interface) Γ(t) which moves inward as time progresses. In other words, the domain Ω(t) expends, that is,Concentration of drug in the liquid phase is given by function p(y, t), and function ϕ(y) describes the drug concentration in the bulk liquid phase. Under conditions of practical significance, the amount of drug loaded is usually much bigger than the solubility of drug, and the time scale for the release process is long. Therefore, the rate of movement of the solid-liquid interface is relatively slow compared to that of drug diffusion.Note that Liu and Xu [29] were the first who introduced the timefractional diffusion equation to the drug release process. They obtained the well-known, semi-empirical formula in the controlled drug release system given by Ritger and Peppas.Problem (1.1)-(1.4) can be called the fractional quasistationary Stefan problem which arises under consideration of the materials with memory [44]. If Eq. (1.1) in problem (1.1)-(1.4) is replaced by the subdiffusion equation, we will get the fractional Stefan problem formulated and studied by Atkinson [1] for the motion of planar, cylindrical and spherical boundaries.If ν = 1, free boundary problem (1.1)-(1.4) is interpreted as the mathematical model of the Hele-Shaw problem, which has a hydrod...