Anomalous solute transport in complex media Abstract
Time-space fractional derivative model
Download 276.08 Kb. Pdf ko'rish
|
7-203-216
- Bu sahifa navigatsiya:
- 2.4 Distributed-order model for multi-scale diffusion
2.3 Time-space fractional derivative model
The time-space fractional derivative model is designed to characterize the solute trans- port process with mixed sub- and superdiffusion due to the competition between the time memory and space non-locality, i. e., 𝜕 α c(x, t) 𝜕 t α = K 𝜕 β c(x, t) 𝜕| x| β . (5) When c(x, 0) = δ(x), the MSD described by model (5) in an unbounded domain can be written as ⟨ x 2 ( t)⟩ ∝ t 2α/β . (6) When 0 < 2α/β < 1, it mainly exhibits subdiffusive behavior, while superdiffusion dominates solute transport when 2α/β > 1. Especially, when α = 1 and β = 2, it reduces to normal diffusion. 2.4 Distributed-order model for multi-scale diffusion Diffusive dynamics in some porous media may evolve when crossing several spatial scales varying from micro-scale to meso-scale and macro-scale. The complex nature, and sometimes non-stationary deposits (due to the change of sediment supply) of the multi-scale heterogeneous medium, can make the single scale description unfeasi- ble. It is necessary to investigate how and to what extent the micro-scale behavior affects the anomalous phenomena observed or measured at macro-scale. Meanwhile, in the time domain, the system may exhibit different behaviors from the short to long time ranges. Hence, the modeling approach for multi-scale diffusion and the cross- scale conveying effect of solute transport have attracted wide attention in the study of anomalous diffusion. The standard fractional derivative models mentioned above fail to depict the dif- fusion processes involving a wide range of time spectra and/or spatial scales. Inspired by the relatively new concept of distributed-order fractional differential modeling, the Anomalous solute transport in complex media | 197 multiple fractional derivative terms can be applied to describe multi-scale diffusion, which leads to the following governing diffusion equation: N ∑ n=1 p(α(n), t)𝜕 α(n) c(x, t) 𝜕 t α(n) = K 𝜕 2 c(x, t) 𝜕 x 2 , (7) where p(α(n), t) is the weight parameter and α(n) ∈ (0, 1) is the order of the time- fractional derivative which characterizes different degrees of subdiffusion. The multi-term operator in (7) corresponds to different diffusive features at dif- ferent time ranges. For instance, the subdiffusive feature may be measured at the micro-second scale; meanwhile, the normal diffusion behavior is observed at the hour scale. Using this model, different diffusion features at different time scales can be captured by one fractional differential equation model, and the coupling effect of multiple scales can also be reflected. This type of model can also be conveniently expanded into the Fokker–Planck equation which involves the convection of contam- inants and can be stated as the following form: N ∑ n=1 p(α(n), t)𝜕 α(n) c(x, t) 𝜕 t α(n) = − A𝜕c(x, t) 𝜕 x + K 𝜕 2 c(x, t) 𝜕 x 2 , (8) where A is the convection coefficient. In addition, when characterizing the multi-scale superdiffusion process, we can use the following multi-term spatial fractional equation as the governing equation: 𝜕 c(x, t) 𝜕 t = N ∑ n=1 p(x, t, β(n))A n 𝜕 2β(n) c(x, t) 𝜕| x| 2β(n) , 0 ≤ β(n) ≤ 1, (9) where β(n) is the order of the spatial fractional derivative, p(x, t, β(n)) denotes the weight parameter, and A n represents the convection or diffusion coefficient. The right- hand side of (9) may include the convection term 𝜕 c(x,t) 𝜕 x , the normal diffusion term 𝜕 2 c(x,t) 𝜕 x 2 , and/or the source term related to c(x, t). The standard spatial fractional dif- fusion model (with a single space-fractional derivative) is designed to characterize superdiffusion, and hence model (9) may depict multi-scale diffusion which involves both superdiffusion and normal diffusion. For instance, the feature of anomalous dif- fusion may be measured at the molecular scale, while the normal diffusion behavior is observed at a larger scale. In other words, different anomalous dynamics dominate at different time ranges, and the competition and coupling of multiple subdiffusion and normal diffusion described by model (9) characterize the scaling behavior well known in hydrological processes. In addition, if we replace the space-fractional term 𝜕 2β(n) c(x,t) 𝜕| x| 2β(n) in the right-hand side of (9) by the fractional Laplacian (−Δ) β(n) c(x, t), it can depict the multi-scale and multi-dimensional diffusion. 198 | H. G. Sun et al. It should be pointed out that, in the above models, if one of the weight parameters is much larger than the others (e. g., p(x, t, α(n)) ≫ p(x, t, α(j)) with j = 1, 2, . . . , n−1, n+ 1, . . . , N), then the above models may reduce to the diffusion models at the single scale. It means that the weighted fractional derivative terms dominate the overall dynamics of multi-scale diffusion. Model (7) has the following solution in the Fourier–Laplace domain: c(k, s) = ∑ N n=1 p(α(n))s α(n)−1 Kk 2 + ∑ N n=1 p(α(n))s α(n) . (10) At the early and late times, the MSD for plumes described by model (7) grows as { ⟨ x 2 ( t)⟩ ∝ t α max , t → 0, ⟨ x 2 ( t)⟩ ∝ t α min , t → ∞, (11) where α max is the maximum order of the space-fractional derivative α(⋅) and α min is the minimum of α(⋅). If N = 1 and α(1) = 1 in (7), it reduces to the MSD of normal diffusion: ⟨ x 2 ( t)⟩ = 2Kt. Different orders of the time-fractional derivative represent different scaling laws, and hence the multi-term model (7) leads to multiple power-law phenomena, where the power-law scaling changes with time. For example, when α max = 1, model (7) cap- tures the transient diffusion which shifts from normal diffusion to subdiffusion with the evolution of time. Download 276.08 Kb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling