Peter E. Malin
Professor & Director
Institute of Earth Science and Engineering
University of Auckland, NZ

will give a talk
on

"Making the Most of a Frac:
What Microearthquake, Electromagnetic, and Emission Tomography Observations in Complex Rock Reveal about Fluid Movements During Permeability Stimulations"

All people interested are invited!

Abstract:

Making the Most of a Frac:
What Microearthquake, Electromagnetic, and Emission Tomography Observations in Complex Rock Reveal about Fluid Movements During Permeability Stimulations

Peter Malin, Institute of Earth Science and Engineering

We have completed a time lapse magnetotellurics (MT) and induced seismicity (MEQ) study of reservoir stimulation at Paralana, South Australia. Unlike induced seismic events, which we originally thought correlate mostly to fluid movement and not injection volume, we anticipated that time lapse MT would be more sensitive to fluid intrusion location and amount. The challenge was detect these two physically independent signals and then combine them into a more revealing image of the stimulation. Further, in separate developments, while still rarely done, Seismic Emission Tomography (SET) observations of both ambient and induced tremors seem to suggest the MEQ data are not in fact strongly coupled to fluid movement.
We deployed at the Paralana stimulation site a dozen borehole MEQ stations, 11 permanent MT stations, and conducted MT surveys before-and-after the stimulation. Approximately 3M liters of brine were injected at ~3.6 km depth, producing over 12,000 locatable seismic events. A correlation was observed between the induced seismicity and very small changes in the MT response.
MT response depends on bulk electrical conductivities and, to a lesser degree, specific locations of conductive fluids. Forward modelling in 3D shows that difference in MT response with and without localized fluids is small: a few degrees in MT phase. Thus, for use in stimulation monitoring it is essential to obtain high resolution, low noise data to produce reliable views of subsurface changes. As a demonstration of this technique, we show and model MT measurements spanning the fluid injection at Paralana. The results indicate MT is capable of imaging locations of injected fluids, with clearer results for shallower injection and/or less conductive cover.
The main confounding issue in the resulting model is the individual signal responses to the inherent heterogeneity of the earth’s crust. The top kms of the earth contain its most rapid changes in resistivity and seismic velocities. The complex and stochastic character of this heterogeneity can be seen in well logs and cores, the spectral decomposition of which reveals their “~1/k” – “pink noise” distribution with distance.
This environment unavoidably confounds accurate and relative location of seismic events. Source and receiver position changes of only tens of meters results in significantly different travel times and waveforms. Consequently, standard location methods produce “clouds” of hypocentres, almost irrespective of where recordings were taken. While it might seem that the long wavelength (diffusion) scales of the MT observations would be more likely to average over such complexity, their 1/k nature also confounds a simplistic analysis of these data.
To add to the confusion, sites of ambient and induced SET activity do not appear directly correlated with MEQ locations. One suggestion is that SET signals relate more directly to fluid movement while MEQ events reveal more about material strengths. While not strongly or quickly resolving these issues, the fact that we are now able to use more than one line of observational investigation on them is a very encouraging development.