Abstract
Hydraulic fracturing has become increasingly prevalent in the oil and gas industry as well as in enhanced geothermal systems as a method to increase the permeability of rock masses. However, the process is still not very well understood in terms of the fracture mechanisms that occur as cracks initiate, propagate and coalesce. Laboratory tests can be useful in this regard as they provide insights into fracture behaviour during hydraulic loading under controlled conditions. This paper describes the development of an acoustic emissions setup capable of capturing a stream of data over approximately five seconds corresponding to crack initiation and propagation.
The AE setup was used in a series of hydraulic fracturing experiments on granite specimens. It was found that for one particular specimen, few emissions were observed until the applied water pressure reached 90% of the failure pressure. After this point it was observed that low energy emissions were produced at a constant rate until approximately 0.1s before failure, whereupon a series of higher energy emissions occur. These higher energy emissions were compared to high speed video taken at 15 000 fps, and it was observed that significant local white patching (microcracking) occurred in this time period. Finally, crack initiation appears to be related to the onset of a high amplitude emission with a duration of 0.3s, as observed using high speed video.
1. INTRODUCTION
It is a well-known phenomenon that the rate of acoustic emissions increases close to the time at which shear failure occurs [1]. This is particularly true in brittle materials, where the rate of emissions often exceeds that which can be handled by the acquisition system [2], [3].
AE systems have traditionally had issues recording data at the time of a fracturing event due to hit based architecture, where the system acquires data in discrete segments and requires time to re-arm after every hit. It is possible to circumvent this problem by use of a continuous waveform recorder [4], however in such an approach the data only span a limited time window due to memory limitations. Consequently, any longer experiment would require a separate hit based acquisition system to acquire data in the early phase of the experiment. The main disadvantages of such a setup is that splitting the sensor signal into two acquisition systems increases the electronic noise in all voltage readings, the continuous data can often consume significant memory space, and analysis can become more difficult due to sheer volume of data. As a result, it would be ideal to develop a single acquisition system with both hit-based and continuous modes of operation.
