Abstract
Hydraulic fracture simulation is a viable tool for optimizing treatments in the field. Current large scale developments create a need for more efficient modeling approaches, which are capable of simulating pad-scale projects. One way to boost computational efficiency of a hydraulic fracturing simulator is to use a coarser mesh. This, however, can noticeably affect accuracy. Traditionally, hydraulic fracturing simulators incrementally propagate fracture by adding one element at a time or by breaking a bond to effectively create a fracture element. The former is the case for displacement discontinuity based methods, while the latter corresponds to finite element and discrete element methods. In this situation, the fracture geometry is inherently quantified within the error bound of a single element size. Alternatively, to reduce this error, one may employ a front tracking algorithm, in which the fracture position varies continuously as a function of the fill of the element. To better understand potential benefits, the purpose of this study is to evaluate accuracy of two hydraulic fracture front algorithms, namely the one with Multi Layer Tip Elements (MuLTipEl) and Implicit Level Set Algorithm (ILSA). Both of these algorithms use the tip asymptotic solution to advance the fracture front, but use very different logic underneath. A series of benchmarking numerical examples with various meshes and the degree of complexity is performed to reveal advantages and limitations of these approaches.
1 Introduction
Hydraulic fracture models are often used to help understanding and optimizing processes occurring in the underground formations. With the development of modern shale plays, the problem scale increased dramatically. Numerous wells are typically drilled within a pad, each of which is then hydraulically fractured with tens of stages. This puts enormous pressure on fracture modeling to become more efficient and allowing to simulate large scale problems. To address the issue, one possibility is to use a coarser mesh, which clearly boosts computational performance. This, however, often leads to inaccurate results. An alternative is to use more sophisticated models that are able to deliver relatively accurate results even on a coarse mesh. In the context of hydraulic fracture modeling, this involves fracture front tracking algorithms as they allow to more accurately describe geometry of hydraulic fractures. The aim of this paper is therefore to evaluate and compare performance of two available fracture front tracking algorithms with respect to accuracy for several benchmarking cases.
