Particle velocity based universal algorithm for numerical simulation of hydraulic fractures

TL;DR

This paper introduces a universal numerical algorithm for hydraulic fracture simulation that employs a novel reduced particle velocity variable, enabling accurate, efficient, and model-independent fracture front tracking across various regimes.

Contribution

The paper presents a new mathematical formulation and a modular numerical scheme utilizing reduced particle velocity for universal hydraulic fracture modeling, applicable to multiple elasticity and propagation regimes.

Findings

Accurate fracture front tracking using explicit formulas.

Robust performance across classical 1D models (PKN and KGD).

Solution verified against analytical benchmarks.

Abstract

In the paper, we propose a new effective mathematical formulation and resulting universal numerical algorithm capable of tackling various HF models in the framework of a unified approach. The presented numerical scheme is not limited to any particular elasticity model or crack propagation regime. Its basic assumptions are: i) proper choice of independent and dependent variables (with the direct utilization of a new one - the reduced particle velocity), ii) tracing the fracture front by use of the speed equation which can be integrated in a closed form and sets an explicit relation between the crack propagation speed and the coefficients in the asymptotic expansion of the crack opening, iii) proper regularization techniques, iv) improved temporal approximation, v) modular algorithm architecture. The application of the new dependent variable, the reduced particle velocity, instead of the usual fluid flow rate, facilitates the computation of the crack propagation speed from the local relation based on the speed equation. As a result, the position of the crack front is accurately determined from an explicit formula derived from the speed equation. The underlying ideas employed in the algorithm are combined together producing a robust and efficient numerical scheme. Its performance is demonstrated using classical examples of 1D models for hydraulic fracturing: PKN and KGD under various fracture propagation regimes. Solution accuracy is verified against dedicated analytical benchmarks and other solutions available in the literature. Most of the ideas developed here, can be directly extended to more general 2D and 3D cases.