Modern theory of hydraulic fracture modeling with using explicit and implicit schemes

05/08/2019 ∙ by A. M. Linkov, et al. ∙ 0

The paper presents novel results, obtained on the basis of the modified theory of hydraulic fractures (HF). The theory underlines significance of the speed equation. When applied to numerical simulation of HF, the theory reveals three distinct issues: (i) modeling the central part of a HF; (ii) modeling the near-front zone; and (iii) tracing changes in the shape of a fracture contour. Modeling the central part leads to a stiff system of ODE in time, what strongly complicates its integration. For explicit schemes, it requires small time steps to meet the CFL condition. For implicit schemes, it requires proper preconditioners. The gains and flaws of the two strategies are discussed. It is noted that a rough spatial mesh may be used in the central part. Modeling the near-front zone reveals the vital role of the speed equation for HF modeling by any method. Its asymptotic analysis has resulted in the fundamental concept of the universal asymptotic umbrella. For the near-front zone, it is also established that a notable part of the zone adjacent to the front propagates virtually as a simple wave. This implies that the CFL condition of stability for this zone is much less restrictive than for the central part of the fracture. Of essence is also that the wave-like propagation of the near front zone makes preferable upwind schemes of time stepping. On whole, the analysis implies that explicit time stepping may be complementing, competitive and even superior over implicit integration. Tracing changes in the front shape appears merely in 3D problems when the contour changes its form. The problem, being essentially geometrical, it may be solved separately by various methods, including fast marching, level set and the simplest string/marker methods. In 1D cases, it does not arise at all. Quantitative estimations and numerical examples illustrate the theoretical conclusions.



There are no comments yet.


This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.