2014_Pittsburgh_Geomechnics_Workshop_final.pdf
Transcript of 2014_Pittsburgh_Geomechnics_Workshop_final.pdf
![Page 1: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/1.jpg)
Recent Developments in the Generalized Finite Element Method for 3-D Hydraulic Fracture Propagation and Interactions C. Armando Duarte Dept. of Civil and Environmental Engineering Computational Science and Engineering
Workshop on Computational Geomechanics University of Pittsburgh, May 22-23, 2014
![Page 2: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/2.jpg)
Objectives • Computational simulations will lead to better designs of hydraulic fracture
treatments, thus reducing the amount of toxic fluids used • Realistic modeling of hydraulic fracturing treatments can, e.g., evaluate the
potential impact of interactions between hydraulic fractures and naturally existing fractures in shale reservoirs
Hydraulic Fracturing of Gas Shale Reservoirs
Motivation • Natural gas production in the US has increased
significantly in the past few years thanks to advances in hydraulic fracturing of gas shale reservoirs
• Yet there are concerns about the environmental impact of toxic fluids used in this process
![Page 3: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/3.jpg)
Acknowledgements
Piyush Gupta*, Jorge Garzon§, Patrick O’Hara¶, Varun Gupta* *University of Illinois at Urbana Champaign, Dept. of Civil and Environmental Engineering, §ExxonMobil Upstream Research Company, ¶Air Force Research Laboratory, Dayton, OH.
![Page 4: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/4.jpg)
What is Hydraulic Fracturing?
Video
Graham Roberts, New York Times, http://www.nytimes.com/interactive/2011/02/27/us/fracking.html 4
![Page 5: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/5.jpg)
Hydraulic Fracturing Simulation
Current Focus: 3-D effects not captured by available simulators • Initial stages of fracture propagation: Fracture re-orientation, interaction and
coalescence
Strategy: Generalized Finite Element Methods
![Page 6: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/6.jpg)
Modeling 3-D Fractures: Limitations of Standard FEM
n It is not “just” fitting the 3-D evolving fracture n FEM meshes must satisfy special requirements for acceptable accuracy
Mesh with quarter-point elements
FEM mesh for a surface fracture
6
![Page 7: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/7.jpg)
Limitations of Standard FEM
• Not possible in general to automatically create structured meshes along both fracture fronts when they are in close proximity
?
7
• Difficulties arise if fracture front is close to complex geometrical features • Fracture surfaces with sharp turns • Coalescence of fractures
• Even with these crafted meshes and quarter-point elements, convergence rate of std FEM is slow (controlled by singularity at fracture front)
![Page 8: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/8.jpg)
Outline
n Motivation and limitations of existing methods
n Basic ideas of GFEM
n GFEM for 3D hydraulic fractures
n Applications
ü Verification
ü Fracture re-orientation
ü Coalescence of 3-D fractures
n Conclusions
![Page 9: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/9.jpg)
Early Works on Generalized FEMs
n Babuska, Caloz and Osborn, 1994 (Special FEM). n Duarte and Oden, 1995 (Hp Clouds). n Babuska and Melenk, 1995 (PUFEM). n Oden, Duarte and Zienkiewicz, 1996 (Hp Clouds/GFEM). n Duarte, Babuska and Oden, 1998 (GFEM). n Belytschko et al., 1999 (Extended FEM). n Strouboulis, Babuska and Copps, 2000 (GFEM).
• Basic idea:
• Use a partition of unity to build Finite Element shape functions
• Review paper Belytschko T., Gracie R. and Ventura G. A review of extended/generalized finite element methods for material modeling, Mod. Simul. Matl. Sci. Eng., 2009 “The XFEM and GFEM are basically identical methods: the name generalized finite element method was adopted by the Texas school in 1995–1996 and the name extended finite element method was coined by the Northwestern school in 1999.”
![Page 10: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/10.jpg)
Generalized Finite Element Method
• GFEM is a Galerkin method with special test/trial space given by
SGFEM = SFEM + SENR
Low order FEM space Enrichment space with functions related to the given problem
L↵i 2 �↵(!↵)
Enrichment function Patch space
SFEM =X
↵2Ih
c↵'↵, c↵ 2 R
SENR =X
↵2Ieh⇢Ih
'↵�↵; �↵ = span{L↵i}m↵i=1
ϕ
1
X
Y
hωα
xα
α
α
![Page 11: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/11.jpg)
Generalized Finite Element Method
• Allows construction of shape functions incorporating a-priori knowledge about solution
Discontinuous enrichment [Moes et al., 1999]
αω
Linear FE shape function
Enrichment function
GFEM shape function
[Oden, Duarte & Zienkiewicz, 1996]
X
↵
'↵(x) = 1
SENR =X
↵2Ieh⇢Ih
'↵�↵; �↵ = span{L↵i}m↵i=1
�↵i(x) = '↵(x)L↵i(x)
![Page 12: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/12.jpg)
A-Priori Error Estimate for the GFEM
![Page 13: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/13.jpg)
GFEM Approximation for 3-D Fractures
[Duarte & Oden 1996]
patch !↵
SGFEM (⌦) =
8><
>:u
hp =X
↵2Ih
'↵(x)| {z }PoU
2
64 u↵(x)| {z }polynomial
+ Hu↵(x)| {z }discontinuous
+ u↵(x)| {z }singular
3
75
9>=
>;
![Page 14: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/14.jpg)
Modeling Fractures with the GFEM
• Discontinuities modeled via enrichment functions, not the FEM mesh • Mesh refinement still required for acceptable accuracy
" = Nodes with discontinuous enrichments Von Mises stress
[Duarte et al., International Journal Numerical Methods in Engineering, 2007]
hp-GFEM
![Page 15: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/15.jpg)
3D Fracture Surface Representation
n High-fidelity explicit representation of fracture surfaces [Duarte et al., 2001, 2009]
n Coalescence of fractures [Garzon et al., 2014]
15
![Page 16: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/16.jpg)
Conditioning of GFEM Approximations
![Page 17: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/17.jpg)
SGFEM: Stable Generalized FEM
![Page 18: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/18.jpg)
SGFEM: Stable Generalized FEM
Linear FE Shape Function
GFEM Enrichment Function
SGFEM Enrichment Function
SGFEM Shape Function GFEM Shape Function
L↵i(x) = L↵i(x)� I!↵(L↵i)(x)
�↵i(x) = '↵(x)L↵i(x)
![Page 19: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/19.jpg)
SGFEM: Stable Generalized FEM
[Gupta, Duarte, Babuska & Banerjee CMAME, 2013]
Conditioning of GFEM/XFEM stiffness matrix O(h�4
)
Conditioning of SGFEM and FEM stiffness matrix O(h�2
)
![Page 20: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/20.jpg)
Selection of Enrichment Functions: Hydraulic Fracturing Regimes
• Fracture propagation is governed by • two competing energy dissipation mechanisms: Viscous flow and fracturing process; • two competing storage mechanisms: In the fracture and in the porous matrix
Hydraulic fracture parametric space* Current Focus: Storage-toughness dominated regime
• Low permeability reservoirs: Neglect flow of hydraulic fluid across fracture faces: • Storage dominated regime
• High confining stress and low viscosity fluid (water): • Constant pressure distribution in fracture; Toughness dominated regime
• Brittle elastic material
Dimensionless toughness
Leak-off coefficient
*[Carrier & Granet, EFM, 2013]
K =4KIcp
⇡
✓1
3Q0E03µ
◆1/4
C = 2CL
✓E0t
12µQ30
◆1/6
![Page 21: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/21.jpg)
Solution for Toughness-Dominated Problem
K = 3
C = 0
Normalized pressure at several time steps
• Solution of coupled problem [Gordeliy & Peirce, cmame, 2013]
Initial fracture
C = 2CL
✓E0t
12µQ30
◆1/6
K =4KIcp
⇡
✓1
3Q0E03µ
◆1/4
![Page 22: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/22.jpg)
Selection of Enrichment Functions: Hydraulic Fracturing Regimes
[Duarte & Oden 1996]
patch !↵
SGFEM (⌦) =
8><
>:u
hp =X
↵2Ih
'↵(x)| {z }PoU
2
64 u↵(x)| {z }polynomial
+ Hu↵(x)| {z }discontinuous
+ u↵(x)| {z }singular
3
75
9>=
>;
Enrichments for toughness-dominated regime:
Valid for toughness-dominated problems
![Page 23: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/23.jpg)
Weak Form at Propagation Step k
Cross section of fracture
![Page 24: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/24.jpg)
Outline
n Motivation and limitations of existing methods
n Basic ideas of GFEM
n GFEM for 3D hydraulic fractures
n Applications
ü Verification
ü Fracture re-orientation
ü Coalescence of 3-D fractures
n Conclusions
![Page 25: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/25.jpg)
a
Verification: Propagation of Circular Fracture*
Geometrical and Computational fracture surface loaded with fluid pressure p
p
*[Gupta & Duarte, 2014]
![Page 26: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/26.jpg)
a
Verification: Propagation of Circular Fracture
Critical pressure
Adopt [Bourdin et al. 2012]:
pc(a) =
✓E⇤Gc⇡
4a
◆1/2
E⇤ = 1
Gc = 1.91⇥ 10�9
a = 0.5
pc(0.5) = 5.477⇥ 10�5
![Page 27: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/27.jpg)
a
Propagation of Circular Fracture
GFEM Model
hmin
/a = 0.016
hmax
/a = 0.027
p-order = 2
N = 215 376 dofs
T = 5.25 min
Critical pressure
phc (0.5) = 5.415⇥ 10�5
er(pc) = 1.15%
phc (a) =Kc
K(a)p
![Page 28: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/28.jpg)
Propagation of Circular Fracture
Repeating for each step of fracture propagation
![Page 29: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/29.jpg)
Application: Fracture Re-Orientation*
• Fracture starts in a direction not perpendicular to minimum in-situ stress • Misalignment of fracture and confining in-situ stresses
a = 10m b = 5m h = 15m p = 3.5 MPa
33 *[Rungamornrat et al., 2005; Gupta & Duarte, 2014]
![Page 30: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/30.jpg)
Fracture Re-Orientation
2a
2b
45o
![Page 31: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/31.jpg)
Fracture Re-Orientation
![Page 32: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/32.jpg)
Fracture Re-Orientation: Step 20
![Page 33: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/33.jpg)
Fracture Re-Orientation: Adaptive Mesh
40
a) b)
Side view Front view
36.4m
13.2m
• Adaptive refinement along fracture front • Sharp features are preserved • High fidelity of fracture surface, regardless of computational mesh
![Page 34: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/34.jpg)
Typical Hydraulic Fracturing
[Z. Rahim et al., 2012]
![Page 35: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/35.jpg)
Longitudinal Fractures
• Develop perpendicular to minimum in-situ stress • Fractures along the length of the wellbore • Planar fractures from the perforation
Perforations
Hydraulic Fracture
![Page 36: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/36.jpg)
Coalescence of Longitudinal Fractures
Challenges
• Propagation and coalescence of multiple fractures • Highly non-convex fracture front after coalescence
Region of coalescence
![Page 37: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/37.jpg)
Coalescence of Longitudinal Fractures
• Propagation and coalescence from a horizontal well
46
h = 2 m p = 3.5 MPa
![Page 38: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/38.jpg)
Coalescence of 3-D Fractures: GFEM Model
• Input mesh and fracture surfaces for GFEM simulation
![Page 39: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/39.jpg)
Coalescence of 3-D Fractures
![Page 40: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/40.jpg)
Coalescence of 3-D Fractures
![Page 41: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/41.jpg)
Coalescence of 3-D Fractures
Fractures just prior to coalescence Fractures just after coalescence
![Page 42: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/42.jpg)
Coalescence of 3-D Fractures
![Page 43: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/43.jpg)
Coalescence of 3-D Fractures
• Coalesced fracture at end of simulation
![Page 44: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/44.jpg)
Coalescence of 3-D Fractures
• Adaptive refinement along fracture fronts
![Page 45: 2014_Pittsburgh_Geomechnics_Workshop_final.pdf](https://reader033.fdocuments.us/reader033/viewer/2022042821/563dbab7550346aa9aa7786c/html5/thumbnails/45.jpg)
Conclusions
n Generalized/Extended FEM removes several limitations of FEM
n It enables the solution of problems that are difficult or not practical
with the FEM
n This is the case of three-dimensional fracture problems involving
ü Complex crack surfaces
ü Fluid-induced fracturing
ü Coalescence of 3-D fractures, etc.
n Open issues under investigation include
ü Coupling with fluid flow on fracture
ü Coalescence of non-planar fractures near a wellbore