You are on page 1of 5

Permeability prediction using pressure and microseismic data: a laboratory experiment

Samik Sil*, David Lane, and Jennifer Gabler, ConocoPhillips; Carl Sondergeld and Chandra Rai, University of
Oklahoma

Summary
For this study, we use data acquired by the University of
We describe a method of determining enhanced Oklahoma (OU) petroleum and geological engineering
permeability from hydraulic fracturing. This method department. A detailed description of the data and related
depends on the analysis of the transient pressure after shut- work can be found in Damani et al. (2012) and Damani
in and integrating it with microseismic derived fracture (2013). In this experiment, a tight sandstone (Lyons
geometry. We present an example of permeability sandstone) core is subjected to tri-axial stress conditions. A
prediction using a laboratory data set from the University vertical hole (or well) is drilled up to the middle of the
of Oklahoma. In this laboratory experiment, a tight sample, and water is pumped at constant rate. The water
sandstone sample is hydraulically fractured. Acoustic produces hydraulic fractures and corresponding
emission data during hydraulic fracturing is recorded and microseismic events. Acoustic emission signals from the
sources relocated. Permeability of the sample is measured sample are recorded by sixteen sensors cemented on the
before and after the fracturing. Predicted permeability using surface and permeability is measured before and after
the described method matches reasonably well with the hydraulic fracturing.
laboratory measured permeability of the post fractured
sample. We show that, from this data set, we can successfully
predict the enhanced permeability. Even though this
Introduction method is tested on laboratory data, in the future, we hope
to apply it to field microseismic data.
Although microseismic data is becoming a powerful tool
for monitoring and determining the geometry of induced Theory of predicting permeability from pressure and
hydraulic fractures, its use as a source of additional microseismic data
geomechanical information is currently limited. These
limitations are largely a consequence of issues with In this section, we describe the method of determining
availability and reliability of the required data. An example permeability using pressure and microseismic event
of information that is desirable but difficult to obtain is the location data. The pressure data used here is from borehole
enhanced permeability due to hydraulic fracturing. pressure, measured during hydraulic fracturing at a certain
Recently, however, two methods of permeability prediction depth. It is assumed that the observed pressure changes are
have been developed by Shapiro et al. (2006) and Grechka related to the changes in the treatment pressure and the
et al. (2010), which depend on the precise location of fracturing mechanism. For this theory to be valid, the
microseismic events. source of pressure diffusion (hydraulic fracture) is an
infinite line source located at a certain depth, z, from the
In this paper we describe a novel method of determining pressure measuring point.
permeability by combining borehole pressure and
microseismic data, a method borrowed from earthquake Pressure induced during any hydraulic fracturing procedure
seismology. After any large earthquake, it is common to starts to diffuse after reaching breakdown pressure and after
observe changes in the water level of distant water wells stopping fluid pumping. Pressure diffusion obeys a
(see Sil, 2006). Sometimes, the observed pressure changes standard 1D diffusivity equation which can be written as
in the wells are due to the passage of a surface wave, which (Brodsky et al., 2003):
enhances the permeability of the aquifer by creating new or
clearing filled fractures (see Brodsky et al. 2003). In this
p 2 p
case, from the typical diffusive or transient change in the D 2 0. (1)
water level (or pressure), one can determine the enhanced t z
permeability of the reservoir. We can apply a similar
technique for calculating the enhanced permeability due to The simplest solution for equation (1) is the standard
the hydraulic fracturing, when continuous pressure data Kelvin solution (Brodsky et al., 2003), which can be
(similar to the water level data) and the corresponding written as:
microseismic event locations are available. In this paper we
present a case study where this technique is applied to
laboratory data to estimate the enhanced permeability due p( z, t ) p0 erf ( z / 4Dt ). (2)
to hydraulic fracturing.

2014 SEG DOI http://dx.doi.org/10.1190/segam2014-0244.1


SEG Denver 2014 Annual Meeting Page 2299
Permeability Prediction

In equations (1) and (2), p is the pore pressure, p0 the initial Laboratory experiment from OU
amplitude of the pressure (i.e. breakdown pressure or ISIP),
t is the time after breakdown or shut-in, z is the depth We test our theory of permeability prediction on a data set
where pore pressure is diffusing, and D is the hydraulic acquired at the University of Oklahoma Petroleum and
diffusivity. Based on equation (2) we can define Geological engineering department laboratory. Although
several samples have been described and analyzed, we
z2
, (3) present our results using data from only one sample (viz.
4D ST-4, see Damani, 2013). ST-4 is a cylindrical Lyons
as the characteristic decay time of the pressure. Therefore, sandstone sample, with low insitu porosity (3-8%) and
we can rewrite equation (2), as (Sil, 2006): permeability (2-20 d). A tri-axial stress system is applied
to the sample with 1=2000 psi, 2=750 psi, and 3=500
psi. The cylinder is 138.8 mm long and has a diameter of
p( z, t ) p0 erf ( ). (4) 99.86 mm. A hole is drilled from surface to a depth of 80.5
mm in the sample for pumping water. Figure 1 is a
Now, by carefully selecting the pressure data after schematic of the sample ST-4.
breakdown and after shut-in we can determine from
equation (4). Distance z can be obtained from microseismic
event locations. For a 2D planer fracture, z is equivalent to
the fracture height. Once z and are available, we can
determine hydraulic diffusivity, D, from equation (3). Then
permeability, K, can be determined using the following
relationship (Shapiro et al. 2003):

NK
D . (5)

Here N is a poroelastic modulus (a function of fluid, grain
and dry frame modulus) and is the fluid dynamic
viscosity. N is defined as:
MPd
N , (6)
H
where,
Figure 1: A schematic diagram of the sample ST-4 (From
1
M
/ K ( ) / K g
, (7) Damani et al. 2012). Stress directions and borehole
locations are shown in the figure.
f

Water is pumped through the borehole at a rate of 10 cc per


1 Kd / K g , (8) minute, and borehole pressure is monitored throughout the
process. When pressure reaches 5000 psi, breakdown
occurs in the sample. Pumping continues for
H Pd 2 M , (9) approximately 120 seconds after this. A total of sixteen
geophones are placed around the sample and continuously
monitor acoustic emission events generated from breaking
Pd K d 4 / 3 d . (10) down the sample. A total of twenty five thousand events
are detected in this experiment, of which we have located
Here Kf,d,g are bulk moduli of the fluid, dry frame, and grain eight thousand. Figure 2 shows a map view of the located
material respectively. d and are the frame shear modulus events.
and porosity respectively. Therefore, when all the
parameters are available, one can determine the
permeability, K, of the fractured medium.

2014 SEG DOI http://dx.doi.org/10.1190/segam2014-0244.1


SEG Denver 2014 Annual Meeting Page 2300
Permeability Prediction

suggests the main hydraulic fracture was created very


quickly as the microseismic events between 100-150s cover
the entire hydraulically fractured zone. A large pumping
induced hydraulic fracture is visible from the surface
(Figure 4) and correlates well with the microseismic events
(Figure 2).

A second burst of events are observed during shut down of


the pumping (before 250s). These events correspond to the
diffusion of induced pressure/fluid without any additional
source. This burst of events during shut-in of the flow is
commonly observed in the field.

A final burst of events (3rd burst) is observed around 350s


of pumping. This burst is due to unloading the applied
stress and can only be observed under laboratory conditions
Figure 2: Map view of the located 8000 microseismic (where the applied stress is physically removed) and may
events from sample ST-4. The main fracture is clearly not be present during hydraulic fracturing in field
visible from the location of these events. Events are conditions. Note that post fractured permeability measured
distributed from the surface through the entire depth of the on the sample should be affected by the presence of the 3 rd
sample. burst of microseismic events. As a result, we expect to
observe higher permeability due to the unloading induced
fracturing.

Figure 3: Top panel: variation of pressure with time (blue


Figure 4: After hydraulic fracturing, the induced fracture is
line).Pumping rate (10 cc/min) is the black broken line.
visible from the top surface (marked with two black lines).
Bottom panel is the r-t plot of the events (see text).
From Damani et al, 2013.
In figure 3, we plot the variation of pressure with time in
the top panel. Pumping rate is also shown here. In the Permeability calculation
bottom panel, we plot the located events distance from the
pumping point as a function of time (r-t plot). This figure To implement the proposed method, we choose the
indicates the event development history with time and their pressure data after breakdown and shut-in. We then fit the
correlation with the different phases of pressure change. selected pressure data to obtain using equation (4). We
use the value of z=139 mm from the microseismic event
From the top panel of Figure 3 we can see that break down location. As discussed in the theory, z in this case (2D
of the sample occurs at around 5000 psi after 100s of fracture) corresponds to the fracture height. Figure 5 shows
pumping. The bottom panel of Figure 3 shows the first the fitting of the data after breakdown (top panel) and after
burst of microseismic events at that time. This figure shut-in (bottom panel).

2014 SEG DOI http://dx.doi.org/10.1190/segam2014-0244.1


SEG Denver 2014 Annual Meeting Page 2301
Permeability Prediction

From the fitting and event locations we obtain D values A similar method of transient pressure analysis to calculate
after breakdown of 7.6X10-4 m2/s and after shut-in of permeability is frequently employed during the pre-fracture
0.002m2/s. From the physical properties of the sample testing phase. Due to the lack of microseismic events at that
(=0.03, Kd=29 GPa, d=26 GPa, Kf = 1.4 GPa, =0.05 Pa phase, however, fracture geometry is always assumed (a
s; see Damani, 2013) and using equation (5), we obtain tuning parameter) for the permeability calculation.
permeability, K, after breakdown of 0.98 md and after shut-
in of 2.6 md. The average value of K measured in the Although in the present study we have obtained the
laboratory after complete fracturing of the sample is 5-6 md permeability values in the range of the observed post-
(which is expected to be higher than the true value due to fracture permeability, this method may contain
later breakdown of the sample by applied stress removal or uncertainties from fracture geometry estimation,
unloading of the stress). microseismic event relocation errors, and/or from the
poroelastic modulus calculation (N). On the other hand, the
estimation of is relatively straight forward. Therefore, one
can compare the value of before (during testing) and after
shut-in to determine the improvement in permeability at a
particular stage by hydraulic fracturing. This can help to
estimate the efficiency of fracturing for that stage.
Comparison of the efficiency among the stages may
correlate with the stage production.

Finally, our method of permeability estimation using


pressure and microseismic data is relatively new, and we
are hopeful it will open a new horizon for research and
collaboration in the near future.

Acknowledgements

We would like to thank ConocoPhillips for allowing us to


present this work. We are thankful to the University of
Oklahoma for performing this experiment and acquiring
acoustic emission data in their lab and sharing the data with
us. Special thanks to Prof. Emily Brodsky for her helpful
comments and suggestions.

Figure 5: Fit of the pressure data after break down (top


panel) and shut-in (bottom panel) with equation (4). Red
dots are the data and the blue line is the model.

Discussion and conclusions

We demonstrated a novel technique of permeability


calculation using pressure and microseismic event location
data from a laboratory experiment.

2014 SEG DOI http://dx.doi.org/10.1190/segam2014-0244.1


SEG Denver 2014 Annual Meeting Page 2302
http://dx.doi.org/10.1190/segam2014-0244.1

EDITED REFERENCES
Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2014
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for
each paper will achieve a high degree of linking to cited sources that appear on the Web.

REFERENCES
Brodsky, E. E., E. Roeloffs, D. Woodcock, I. Gall, and M. Manga, 2003, A mechanism for sustained
groundwater pressure changes induced by distant earthquakes: Journal of Geophysical Research:
Solid Earth, 108, no. B8, http://dx.doi.org/10.1029/2002JB002321.
Damani, A., 2013, Acoustic mapping and fractography of laboratory induced hydraulic fractures: M. S.
thesis, University of Oklahoma.
Damani, A., A. Sharma, C. H. Sondergeld, and C. S. Rai, 2012, Mapping of hydraulic fractures under
triaxial stress conditions in laboratory experiments using acoustic emissions: SPE Annual Technical
Conference and Exhibition, SPE 159604.
Grechka, V., P. Mazumdar, and S. A. Shapiro, 2010, Predicting permeability and gas production of
hydraulically fractured tight sands from microseismic data: Geophysics, 75, no. 1, B1B10,
http://dx.doi.org/10.1190/1.3278724.
Shapiro, S. A., C. Dinske, and E. Rothert, 2006, Hydraulic -fracturing controlled dynamics of
microseismic clouds: Geophysical Research Letters, 33, no. 14, L14312,
http://dx.doi.org/10.1029/2006GL026365.
Sil, S., 2006, Response of Alaskan wells to near and distant large earthquakes: M.S. thesis, University of
Alaska Fairbanks.

2014 SEG DOI http://dx.doi.org/10.1190/segam2014-0244.1


SEG Denver 2014 Annual Meeting Page 2303

You might also like