ABSTRACT

INTRODUCTION

Hydraulic fracturing is widely practiced for well stimulation and stress measurement in the petroleum and rock engineering field. A considerable amount of research has been carried out in the past few decades on this rather simple process of inducing fractures in geologic formation, and brought about a vast improvement in the understanding of the mechanics of hydraulic fracturing. A variety of hydraulic fracturing simulators, primarily numerical models, have been developed to aid in the design of field operations and to provide services to field engineers. These models, however, do not take into full account spatial variations of strength and deformation properties, the effect of the fracture process zone, stress contrast and many other factors, which contribute to the discrepancies between field observations and predictions based on numerical and analytical methods [1, 2].

Invariably these models predict the fractures to initiate in a direction perpendicular to the least principal stress and remain planar as they propagate. Mineback experiments [3] and laboratory test results [1] indicate that the conventional belief of planar fractures are not entirely valid, and also that the effect of the fracture process zone should be taken into consideration in the prediction of fracture growth. Vinegar et al [4] observed a fracture process zone as wide as 15 m.

In our investigation we attempted to study the effects of mechanical property variations and in situ stress contrasts on fracture initiation and on propagation, and the process zone development during hydraulic fracturing. As a first step we conducted parametric analyses using the Dynamic Network Model [5] to gain an insight on the hydraulically driven fracture growth phenomena in rock. The model consists of a network of particles and springs interconnected to each other. Particle masses provide the inertial effect, and spring constants represent the elastic properties of rock. If a spring is stretched beyond a prescribed threshold, the spring breaks to initiate a microfracture. The stresses will then be dynamically redistributed to the rest of the system, affecting the fracture growth thereafter. The model is distinctly different from conventional numerical methods used for fracture process simulation and presents an

opportunity of solving various rock engineering problems.

A series of parametric analyses were carried out to simulate typical cases of hydraulic fracturing in hard brittle rocks. The results indicate that fractures bifurcate as they meet weak links along the path of their progression. This results in the formation of fracture process zones which have similar appearances to those experimentally observed previously. More than two main fractures develop in some cases such as high material properties variation and low contrast between minimum and maximum stresses. These initial results demonstrate clearly that material properties and in situ stress variations have a significant effect on fracture growth during the hydraulic fracturing process. This modeling exercise sheds lights on the fracture initiation and propagation processes and indicate that a further improvement of the model could enhance substantially the current state of understanding of hydraulic fracturing.

This content is only available via PDF.
You can access this article if you purchase or spend a download.