Sie sind auf Seite 1von 10

SPE-189895-MS

Proppant Transport in Complex Fracture Networks

Kaustubh Shrivastava and Mukul M. Sharma, The University of Texas at Austin

Copyright 2018, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Hydraulic Fracturing Technology Conference & Exhibition held in The Woodlands, Texas, USA, 23-25 January 2018.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
During hydraulic fracturing, natural fractures and bedding planes can intersect with growing hydraulic
fractures and form complex fracture networks. This can result in the flow of fluid and proppant in convoluted
fracture pathways with highly variable fracture width and height. Existing models of hydraulic fracturing
assume a planar fracture geometry and are unable to simulate proppant placement in such complex networks.
In this work, we investigate proppant transport in growing fracture networks using a fully three-dimensional,
geomechanical fracture flow, network model with the ability to simulate proppant transport.
A three-dimensional hydraulic fracturing simulator developed using the displacement discontinuity
method is coupled with a network model for proppant transport. The simulator captures the effect of
proppant concentration, fracture width, and fluid rheology on proppant transport. The equations for the
fracture network geomechanics, the fluid flow, and the proppant transport are solved in a coupled manner.
This provides an accurate estimation of both the fluid pressure and the proppant distribution as the fracture
network grows. The geometry of each fracture segment affects the flow distribution in the network.
Simulations are then conducted to study the redistribution of proppant as it settles in the fracture network
during shut-in to get the final proppant distribution in the network.
It is observed that changes in the in-situ stress due to heterogeneity and the stress-shadow induced near
the intersection of a hydraulic fracture and a natural fracture may reduce the fracture width and suppress
the ability of the proppant to move into the natural fracture. In low permeability formations, due to low
leak-off rates, the proppant almost always forms a proppant bank at the bottom of the fracture during shut-
in. For planar fractures, proppant settling may disconnect the conductive proppant bank from the wellbore,
isolating the productive propped fracture from the wellbore. This problem is exaggerated in the case of
fracture networks, where every intersection point between fractures can potentially act as a bottleneck for
the flow of produced hydrocarbons. The increase in the surface area due to hydraulically connected natural
fractures increases fluid leak-off, reduces the average width of the fracture network, increases proppant
concentration, and increases the likelihood of proppant bridging.
This work allows us to improve our understanding of proppant placement in three-dimensional,
mechanically interacting, complex fracture networks. By coupling geomechanics with proppant transport
in fracture networks, it is now possible to study the impact of the stress shadow on proppant placement
2 SPE-189895-MS

in natural fractures. The results will assist in improving hydraulic fracture design for naturally fractured
reservoirs.

Introduction
The success of hydraulic fracturing treatments requires the formation of proppant filled conductive pathways
connecting the created fracture area to the wellbore. The goal of the design engineer is to maximize the
area of the pay zone connected to the wellbore with a propped channel. Simulators are often employed
for determining the final distribution of the proppant inside the fracture. The assumptions made during the
development of the model impact the predicted proppant distribution. The models typically assume idealized
single planar fracture propagation which is seldom expected in the field. Several studies have shown that the
fractures can be complex due to their interaction with natural fractures and bedding planes (Cipolla et al.,
2008; Warpinski and Teufel, 1987). A few experimental attempts have been made to understand proppant
transport in fracture networks. These studies are usually conducted with the assistance of Hele-Shaw
like fixed geometries (Sahai et al., 2014)(Tong and Mohanty, 2016). Although these studies improve our
understanding of proppant transport in fracture networks, they are unable to capture the effect of changing
fracture widths caused by changes in stresses and stress interference between fractures, which can play an
important role in proppant distribution. The opening of the hydraulically connected natural fracture in the
vicinity of a hydraulic fracture can result in high stresses near the region of fracture intersection. This can
limit the growth of the fractures and create bottlenecks for the flow of proppant. This behavior can become
much more complicated in three-dimensions, where an elliptical front of the growing hydraulic fracture
intersects with the natural fracture. The location, orientation, height, and the strength of the natural fracture
can significantly influence fracture growth and proppant transport. The flow direction of the proppant and
the fluid is influenced by the direction of the fracture propagation. Also, in low permeability formations due
to the low leak-off rate, the fracture closure can take up to several days. During this period, the proppant
can redistribute inside the fracture and may form a proppant bank at the bottom of the fracture (Warpinski
et al., 2009). This behavior is observed in the simulated cases. This can result in a large area of the fracture
remaining unpropped. Although these unpropped regions can exhibit residual permeability (Wu et al., 2017),
they can severely limit the production of the hydrocarbons.
Proppant transport is a very complicated multi-physics process. The proppant distribution is determined
by the relative magnitude of the various physical processes involved. To understand the behavior of proppant
transport in complex fracture networks, a fully coupled three-dimensional hydraulic fracturing simulator
is developed. The simulator captures geomechanics (stress-shadow), fluid flow, fluid leak-off, proppant
transport, and the interaction of natural fractures and hydraulic fractures.

Methodology
A hydraulic fracturing simulator is developed using a fully three-dimensional displacement discontinuity
method (DDM) to model fracture growth and proppant transport in naturally fractured reservoirs. DDM is
a boundary element method, and hence the problem is solved only on the surface of the fracture domain.
This makes the method particularly attractive for problems with low surface to volume ratio. It obviates the
need to discretize the entire reservoir, as is required in FVM/FEM techniques, and reduces the dimensions
of the problem by one. In this method, the fracture is described as a thin slit, and the fracture surface is
discretized into rectangular grids. The pressure and the width of each grid are assumed to be constant. The
geomechanical relationship between the displacement discontinuity and the stresses on the fracture surface
is linear as shown in equation 1.

(1a)
SPE-189895-MS 3

(1b)

(1c)

where are tractions acting on the ith fracture element surface, are the displacement
discontinuity associated with the j element, N is the total number of elements in the domain, and A are
th

the influence coefficients.


The fracturing fluid in the model is considered to be continuous, slightly compressible, and isothermal.
The fluid is modeled as a mixture of two components, the proppant particles and the base fluid. It is assumed
that the slurry (a mixture of proppant and base fluid) completely fills the fracture, and fluid lag behind the
fracture tip is considered to be negligible. Fluid leak-off is determined using Carter's leak-off equation. The
flow of the slurry inside the fracture is described using the Reynolds lubrication theory as follows,

(2)

where ρ is the density of the slurry, v is the volume of fracture, k is fracture permeability, A is cross sectional
area of flow, P is fluid pressure, μ is fluid viscosity, is the rate at which fluid leaks off into the
formation, and is the fluid injection rate.
A finite difference method is used to solve the fluid flow equation. The fracture is discretized into thin
rectangular grids to describe the fluid flow, with each grid having a different but constant width. Each grid
is assumed to be a thin parallel slit, and its permeability is given by

(3)

where k is the permeability of the fracture, and w is the fracture width.


The fracture's finite difference mesh can be stitched to other meshes (hydraulically connected fractures)
to distribute fluid according to the pressure distribution of the entire network (Figure 1).

Figure 1—Multiple fluid flow meshes stitched together at their point of intersection to form a complex fracture network

The proppant particles are assumed to be in suspension in the fracturing fluid. The relative motion of the
proppant particles with respect to the fluid is considered due to two reasons: 1) Settling due to the difference
in the densities of the fluid and the proppant 2) The retardation due to the slip of the proppant particles
relative to the fluid.
4 SPE-189895-MS

The proppant distribution inside the fracture is governed by the scalar transport equation (Eq. 4). A no-
flux boundary condition is used for the proppant on the fracture surface.

(4)

where w is the fracture width, c is proppant concentration, and v is the proppant velocity.
The proppant distribution equation is solved after each converged solution is obtained from the coupled
fluid flow and geomechanics system of equations. The resultant velocity of the proppant particles is obtained
by a vectorial summation of the proppant velocity due to settling, and the velocity due to the motion of the
base fluid (retardation). The retardation factor for proppant transport is determined using the correlation
developed by Blyton et al. (2015). These correlations provide the retardation factor of the proppant based
on the diameter of the proppant, width of the fracture, concentration of the fracture, and Reynolds number
of the flow. The settling velocity of proppant is calculated from the correlations developed by Gadde et al.
(2004), based on their experimental investigation. In the simulator, proppant transport only due to advection
is considered. A fully implicit finite difference method with an upwinding scheme is used to solve the
proppant transport equation (Eq. 4) in order to avoid numerical dispersion. In addition, the presence of the
suspended solid particles (proppant) in the slurry changes its rheological behavior. Correlations developed
by Shah (1993) are incorporated to capture the effect of proppant particles on the rheology of the slurry.
The governing equation of geomechanics and the fluid flow are solved in a coupled manner to obtain the
fracture pressure and width. Local linearization is employed to improve the speed of convergence for the
coupled system (Shrivastava et al., 2017). The obtained solution is used to compute the stress intensity at the
fracture tip, which is compared with the critical stress intensity of the matrix in order to determine fracture
growth. The fracture is allowed to grow until the stress intensity at the fracture tip becomes lower than the
critical stress intensity. This allows the model to predict the transient dimensions of the growing fracture.
In the model, the interaction between the hydraulic fractures and the natural fractures is accounted for
in two ways: 1) the hydraulic fracture is allowed to merge with natural fractures in case of an intersection
between them based on the crossing criterion given by Wu and Olson (2014); 2) the stresses at hydraulically
disconnected natural fractures are calculated at each time step, and their slippage is predicted based on the
Mohr-Coulomb criterion. The slipped natural fractures are converted into DDM elements which allows the
model to capture the effect of stress relaxation due to the opening of disconnected natural fractures.
During fracture growth, the fracture propagation direction is determined using a maximum
circumferential stress criterion (Erdogan and Sih, 1963). In case of a hydraulic fracture's intersection with
a natural fracture, the fracture tip follows the geometry of the natural fracture as it provides an easier path
for the propagation of the fracture tip (Figure 2).

Figure 2—Intersection of a growing hydraulic fracture with a natural fracture. The fracture propagates along the
natural fracture for its length. It then starts to observe far-field stresses and reorients itself along the Shmax direction.
SPE-189895-MS 5

The stress interference between hydraulic fractures and natural fractures can play an important role
in determining the final dimensions of the fracture (Shrivastava and Sharma, 2018). In order to capture
the stress interaction between all the discontinuous surfaces (hydraulic fractures, hydraulically connected
natural fractures, and disconnected natural fractures), the entire problem is considered to be in a single
domain. This allows us to capture the effect of hydraulic fracture opening of nearby fractures.

Results
Effect of natural fractures on proppant transport
A simulation case is set up to understand the effect of natural fractures on proppant transport. To simulate
complex fracture network creation, a hydraulic fracture is allowed to grow radially from a horizontal well.
Two natural fractures are placed in the simulation domain (Figure 3). When the hydraulic fracture intersects
the natural fracture, the fluid inside the hydraulic fracture flows into the natural fracture. This reduces the
width of the hydraulic fracture and transports proppant and fluid into the natural fracture. The process is
repeated for the next intersection. It is observed that the higher stresses generated due to the opening of the
fractures reduces the width of the elements near the intersection and retard the growth of the fracture. This
results in the formation of a thin channel for fluid and proppant transport between the connected fractures.
This channel may grow if higher growth resistance is encountered at other fracture tips and this region
becomes the path of least resistance available for fracture growth. Proppant transport is simulated during
the fracture growth process to predict the distribution of proppant inside the fracture network (Figure 4).
The parameters used in the simulation are shown in Table 1. At the end of pumping, a uniform proppant
distribution is observed in all the three fractures (Figure 4a). The fracture farthest away from the wellbore
is observed to have the lowest proppant concentration because of proppant retardation.

Figure 3—The geometry of a simulation case to investigate the effect of natural fractures on proppant transport.
6 SPE-189895-MS

Figure 4—Figure 4a shows the distribution of proppant in a complex fracture network generated
due to natural fractures at the end of pumping. 4b shows the final distribution of proppant.
Note that in many cases the proppant loses connectivity with the wellbore after settling.

Table 1—Parameters used for simulation study of fracture network formation.

Parameter Value Units

Young's Modulus 1 Mpsi


Poisson Ratio 0.25
Shmin 3000 psi
Shmax 3060 psi
Injection Rate 5 bbl / min
Pumping Time 60 seconds
Simulation Time 80 minutes
Maximum Proppant Concentration 0.1

Subsequently, the post-pumping behavior of the fracture is simulated, and the proppant is allowed to
settle during fracture closure due to leak-off. It is observed that the proppant settles to the bottom of the
fracture and forms proppant banks in all of the three fractures (Figure 4b). The simulation is not continued
once the change in the proppant distribution becomes insignificant.

Effect of Bedding Planes


A case is set up to study the effect of bedding planes on fracture propagation and proppant distribution
(Figure 5). A radially growing fracture is allowed to intersect a perfectly horizontal weak bedding plane
(a region with lower K1c), which then grows and intersects with a natural fracture. The parameters used
in the simulation are shown in Table 2. The simulation is conducted with a low vertical stress to facilitate
the opening of the bedding plane. The bedding plane slips as the hydraulic fracture approaches, grows
radially and then intersects with a natural fracture. The proppant and the fluid are distributed in the three
hydraulically connected fractures (Figure 6a). The vertical natural fracture grows upwards to avoid the
higher stress region generated due to the opening of the initial hydraulic fracture. Following injection, post-
pumping redistribution of proppant is simulated under shut-in (injection rate = 0) conditions, and the fracture
is allowed to close due to fluid leak-off. It is observed that the bedding plane remains propped during shut-
in and the proppant inside the vertical fracture settles to form proppant banks. These proppant banks remain
disconnected from each other and from the wellbore, which can be detrimental for hydrocarbon production
from the fracture.
SPE-189895-MS 7

Figure 5—The geometry of the simulation case to investigate the effect of bedding planes on proppant transport.

Figure 6—6a shows the distribution of proppant in a complex fracture network generated by the interaction of
the growing hydraulic fracture with a bedding plane. 6b shows the final distribution of proppant for the case.

Table 2—Parameters used for simulation study of fracture network formation due to bedding planes.

Parameter Value Units

Young's Modulus 1 Mpsi


Poisson Ratio 0.25
Shmin 3000 psi
Sv(vertical stress) 3060 psi
Injection Rate 5 bbl / min
Pumping Time 60 seconds
Simulation Time 80 minutes
Maximum Proppant Concentration 0.1
8 SPE-189895-MS

Effect of Proppant Size


An investigation of the effect of proppant size on the transport of proppant in a fracture network is conducted
to see if there is a benefit to using smaller size proppants. Two simulation cases are set up. In the first case,
a 40-70 mesh proppant size is chosen and is compared with the second case which assumes a 100 mesh
proppant size. A hydraulic fracture is allowed to grow radially from a horizontal wellbore until it intersects
with a natural fracture (Figure 7). The parameters used for the simulation for these cases are shown in Table
3. It is observed that for the case of the 40-70 mesh, the proppant starts to accumulate in front of the region
of fracture intersection. This is due to the lower width of the intersection element which is the result of
the higher stress near the fracture intersection. No such accumulation is observed in the case of the 100
mesh proppant size.

Figure 7—Comparison of the distribution of proppant in a fracture network with different proppant sizes.
Fig 7a shows the results for a proppant size of 100#. Fig 7b shows the results for a proppant size of 40-70#

Table 3—Parameters used for investigation of effect of proppant size on proppant transport

Parameter Value Units

Young's Modulus 2.15 Mpsi


Poisson Ratio 0.25
Shmin 3000 psi
Sv(vertical stress) 3060 psi
Injection Rate 5 bbl / min
Pumping Time 400 seconds
Maximum Proppant Concentration 0.1

The lower width increases the retardation of the proppant and leads to an increase in proppant
concentration at the point of fracture intersection. A similar situation will arise if the proppant concentration
is increased, leading to the possibility of tip screen-out. Since multiple fracture tips are propagating this
may not lead to a sharp increase in treating pressure (as would be the case for a single planar fracture).
Such restrictions to proppant transport can also be caused by heterogeneity on the in-situ stresses (caused
by changes in mineralogy). Results obtained from our model for such situations (results not shown) clearly
show that changes in fracture width can result in proppant bridging. In certain situations, the width can
become lower than the size of the proppant and restrict its transport. It is recommended that the design
SPE-189895-MS 9

engineer conduct such simulations to understand the expected width of fracture intersection regions for
proppant size selection.
Since it is not possible to obtain a detailed description of the in-situ natural fracture network or bedding
planes in any given geologic setting, the geometry of the fracture can only be predicted in a statistical
sense. It is essential to obtain a statistical description of the natural fracture system, the in-situ stress and the
bedding planes to describe the producing interval. The ultimate fracture geometry will be a strong function
of these parameters. Changes in fracture design based on simulations are still essential since the simulations
do provide excellent guidance on how the fracture geometry will change as the fracture design parameters
are changed. While the precise geometry of the fracture may not be predicted (due to the uncertainty in
the input geology), the fracture network area and complexity, and how much of the fracture might remain
propped are accurately estimated. Fortunately, the productivity of the well is determined primary by the first
order effects such as fracture area, fracture complexity (orientation and branching) and proppant placement
while the details of the fracture geometry are less important.

Conclusion
• When hydraulic fractures intersect natural fractures and bedding planes they create complex
fracture networks. A model is presented that allows us to model this dynamic process of the growth
of fracture networks and proppant transport in geologically complex formations while accounting
for geomechanical interaction between all the fractures and slippage at weak planes.
• During fracture intersection, the region near the intersection experiences high stresses (caused
by the stress shadow of the fractures) and limits the width of the intersecting fractures. These
intersections may then act as a bottleneck for the flow of proppants.
• In many instances, bedding planes (horizontal and dipping) can slip, dilate and allow proppant
transport. It is important for such bedding planes to remain propped to provide a continuous
conductive channel between vertical fractures.
• Proppant settling during shut-in will result in the formation of proppant banks in vertical fractures.
When this happens, the connectivity of the vertical fractures with the wellbore and with other
fractures is significantly impaired. This can lead to a large loss in well productivity.
• The use of higher mesh size proppant (smaller proppant) for fracturing has clear benefits when
fracturing naturally fractured formations. The ability of the smaller proppant to enter natural
fractures and increase the propped area is essential to improving well productivity and fracture
connectivity. Smaller proppants are more likely to cross fracture intersections into natural fractures
or sheared bedding planes. Larger proppants may experience higher retardation factors and form a
proppant bank at the fracture intersection. The use of smaller proppant sizes is also an advantage
in minimizing settling both during pumping and during shut-in.
• Smaller concentration of proppants should be injected in naturally fractured formations to ensure
that proppant buildup and bridging at fracture intersections does not cause a tip screenout. The
large number of propagating fracture tips helps to divert the fracture as some tips screen out.
• Low permeability formations experience very low leak-off rates and may take several days to close
on the proppant particles. This provides ample time for proppant settling leading to proppant bank
formation at the bottom of the fracture. We observe that this leads to propped regions in the fracture
becoming disconnected from the fracture network and from the wellbore and limiting production.

Acknowledgement
The authors would like to acknowledge the funding and support from the member companies of the
Hydraulic Fracturing and Sand Control Joint Industry Consortium at the University of Texas at Austin.
10 SPE-189895-MS

References
Blyton, C.A.J., Gala, D.P., Sharma, M.M., 2015. A Comprehensive Study of Proppant Transport in a Hydraulic
Fracture. Presented at the SPE Annual Technical Conference and Exhibition, Society of Petroleum Engineers. https://
doi.org/10.2118/174973-MS
Cipolla, C.L., Warpinski, N.R., Mayerhofer, M.J., 2008. Hydraulic Fracture Complexity: Diagnosis, Remediation, And
Exploration. Presented at the SPE Asia Pacific Oil and Gas Conference and Exhibition, Society of Petroleum
Engineers. https://doi.org/10.2118/115771-MS
Erdogan, F., Sih, G.C., 1963. On the Crack Extension in Plates Under Plane Loading and Transverse Shear. J. Basic Eng.
85, 519–525. https://doi.org/10.1115/1.3656897
Gadde, P.B., Liu, Y., Norman, J., Bonnecaze, R., Sharma, M.M., 2004. Modeling Proppant Settling in Water-Fracs.
Presented at the SPE Annual Technical Conference and Exhibition, Society of Petroleum Engineers. https://
doi.org/10.2118/89875-MS
Sahai, R., Miskimins, J.L., Olson, K.E., 2014. Laboratory Results of Proppant Transport in Complex Fracture Systems.
Presented at the SPE Hydraulic Fracturing Technology Conference, Society of Petroleum Engineers. https://
doi.org/10.2118/168579-MS
Shah, S.N., 1993. Rheological Characterization of Hydraulic Fracturing Slurries. SPE Prod. Facil. 8, 123–130. https://
doi.org/10.2118/22839-PA
Shrivastava, K., Blyton, C.A.J., Sharma, M.M., 2017. Local Linearization Method for Efficient Solution of Coupled Fluid
Flow and Geomechanics Problem. Presented at the 51st U.S. Rock Mechanics/Geomechanics Symposium, American
Rock Mechanics Association.
Shrivastava, K., Sharma, M.M., 2018. Mechanisms for the Formation of Complex Fracture Networks in Naturally
Fractured Rocks. Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition, Society of
Petroleum Engineers, Woodlands, Texas.
Tong, S., Mohanty, K.K., 2016. Proppant transport study in fractures with intersections. Fuel 181, 463–477. https://
doi.org/10.1016/j.fuel.2016.04.144
Warpinski, N.R., Mayerhofer, M.J., Vincent, M.C., Cipolla, C.L., Lolon, E.P., 2009. Stimulating Unconventional
Reservoirs: Maximizing Network Growth While Optimizing Fracture Conductivity. J. Can. Pet. Technol. 48, 39–51.
https://doi.org/10.2118/114173-PA
Warpinski, N.R., Teufel, L.W., 1987. Influence of Geologic Discontinuities on Hydraulic Fracture Propagation (includes
associated papers 17011 and 17074). J. Pet. Technol. 39, 209–220. https://doi.org/10.2118/13224-PA
Wu, K., Olson, J.E., 2014. Mechanics Analysis of Interaction Between Hydraulic and Natural Fractures in
Shale Reservoirs. Presented at the Unconventional Resources Technology Conference, Unconventional Resources
Technology Conference. https://doi.org/10.15530/URTEC-2014-1922946
Wu, W., Kakkar, P., Zhou, J., Russell, R., Sharma, M.M., 2017. An Experimental Investigation of the Conductivity of
Unpropped Fractures in Shales. Presented at the SPE Hydraulic Fracturing Technology Conference and Exhibition,
Society of Petroleum Engineers. https://doi.org/10.2118/184858-MS

Das könnte Ihnen auch gefallen