Professional Documents
Culture Documents
Norway
Motivation
CCS currently great interest in how one should go about in order to facilitate Carbon Capture and Storage. Politicians and environmentalists place a lot of emphasis on this. It is the storage of carbon that concerns us in the context of hydraulic fracture.
Hydraulic fracture is a well known method to enhance oil and gas production. Interest in this approach is increasing as the remaining oil becomes more difficult to extract. Also has an environmental side, since a lot of attention has been focused on pollution from oil extraction operations.
Physics. Hydraulic fracture also represents an interesting phenomenon from the academic point of view. Here we can test to what extent our physical understanding of fracture and flow tallies with experimental reality.
Our project
CCS, concerns geologic storage of CO2. A grant from the Norwegian Research Council via their CLIMIT program for the project Efficient CO2 Absorption in Water-Saturated Porous Media through Hydraulic Fracture. This is a joint project between SINTEF and NTNU.
In analogy with this Maurice Biot (1941) introduced a similar quantity for the fluid response, i.e.,
Since this is the relative movement of the fluid with respect to the solid, this can also be stated as
We then get
where we have assumed that the porosity is the same throughout the control volume.
Next, we have for the right-hand side
This results in
Considering all three spatial directions and adding the contributions, we get
10
11
to obtain
12
14
15
16
17
18
19
20
21
On a triangular lattice there are six neighbours to each node. These are numbered from one to six, beginning with the horizontal beam which extends towards the left-hand side.
22
Forces between the nodes are defined IN ANALOGY with ELASTIC BEAMS. The beams are fastened onto each other in such a way that the angle between them (60 degrees) is preserved, even when there is a rotation at the node.
23
We place a COORDINATE SYSTEM on each node in order to keep track of the displacements.
24
The beams are assumed to be brittle-elastic and the elastic response is a combination of AXIAL, SHEAR and FLEXURAL forces. Thick beams have been assumed in the expressions derived.
25
Depending on the combination of displacements at any given point in time, the forces are calculated accordingly..
26
Poroelastic beams
To include Biots theory in our model of hydraulic fracture we include in our beams the assumption that they are porous.
Flow through the beams are along the axis between the nodes that it connects.
27
Poroelastic beams
To include Biots theory in our model of hydraulic fracture we include in our beams the assumption that they are porous.
Flow through the beams are along the axis between the nodes that it connects.
To say a few words about the Biot-Willis parameter which couples in the pore pressure, this expresses how much of the bulk strain is taken up by the change in the pore volume and how much is taken up by the change in the solid volume.
28
Biot-Willis parameter
The governing equation of mechanical equilibrium was obtained from the two constitutive equation resulting from Biots coupling of volumetric strain-stress with increment in fluid content-pore pressure.
The expression for the parameter which obtains in the derivation of mechanical equilibrium in the equation at the top is
29
Biot-Willis parameter
PHYSICAL INTERPRETATION
The Biot-Willis parameter is an expression of the increment in fluid content with respect to changes in volumetric strain. A number between zero and one having the following interpretation:
INCOMPRESSIBLE LIQUID All the bulk strain is taken up by the change in solid volume INCOMPRESSIBLE SOLID All the bulk strain is taken up by the change in pore volume
(Berea sandstone)
30
0.79
Here F is the axial force and M is the bending moment. Random thresholds are generated for each beam on the lattice, one for the amount of axial force that the beam can withstand, and one for the amount of bending it can withstand. These are then combined as shown. When a beam breaks it is removed irreversibly from the lattice, only retaining its axial force where relative displacements indicate local compression. Axial contribution to beam breaking is assumed to occur only when beams are loaded in tension.
Disorder
Thresholds are generated by raising a random number to a power D, a positive power indicates that some thresholds deviate towards weak strength. Likewise, A negative D indicates that some of the thresholds deviate towards stronger strength. The magnitude of D controls how many of the thresholds deviate in this way, i.e., how strong the disorder is.
31
Crack growth and disorder uniaxial loading of a sheet with a central crack
Fracture criterion:
Crack tip: high stress intensity
STRETCHED
No disorder: fracture is stress dominated, and crack growth localized to a single existing crack. Unstable.
Disorder: if the local variation in material strength is strong enough, new cracks will appear randomly. Fracture is disorder dominated. Stable.
COMPRESSED
32
33
Numerical calculation
Fracture in our model is driven by a potential gradient which is set up by injecting fluid at the center of the system. Here the pressure is kept constant throughout the fracturing process.
34
Numerical calculation
Fracture in our model is driven by a potential gradient which is set up by injecting fluid at the center of the system. Here the pressure is kept constant throughout the fracturing process.
CONJUGATE GRADIENTS
Since the expressions which we derive for the poroelastic forces in the system are all linear, the resulting elastic energy expression is quadratic.
35
Numerical calculation
Fracture in our model is driven by a potential gradient which is set up by injecting fluid at the center of the system. Here the pressure is kept constant throughout the fracturing process.
CONJUGATE GRADIENTS
Since the expressions which we derive for the poroelastic forces in the system are all linear, the resulting elastic energy expression is quadratic. An efficient way to minimize a quadratic form is to use conjugate gradients. The minimum in our system corresponds to that situation where on each node the sum of forces and moments is zero. Physically, this is the required situation for a system in mechanical equilibrium.
36
Numerical calculation
Fracture in our model is driven by a potential gradient which is set up by injecting fluid at the center of the system. Here the pressure is kept constant throughout the fracturing process. The permeability in the system is set to be FINITE wherever the system is intact and INFINITE within cracks.
As cracks grow on the lattice the pressure is updated using a CLUSTER MAPPING algorithm which makes sure that the pressure distribution is consistent with our assumption on the permeability.
Each time a crack touches the boundary the pressure within that crack is vented, that is, it is set to zero. The process is continued until one of the cracks which connect to the injection hole reaches the boundary. This is a consistent stopping criterion.
37
N= 0 beams broken
38
39
40
41
42
43
44
Crack Samples
Disorder D=1.4
45
46
Crack Samples
Disorder D=2
47
48
Crack Samples
Disorder D=0.4
49
50
Crack Samples
Disorder D=-4
51
52
53
54
55
Crack Samples
Disorder D=-6
56
57
58
59
60
61
Essentially, this is the ratio of stress to strenght. Its intensity increases as the stress increases and it also increases as the strength decreases.
62
63
D>0
Here we show the average force-pressure responce for 1000 samples each of systems with disorders ranging from D=0.4 (A) to D=2 (H) in steps of 0.2
64
D<0
Force-pressure responce for 1000 samples each of systems with disorders D=-2 (A), D=-3 (B) D=-4 (C) and D=-5 (D)
65
Force-pressure responce for 1000 samples of D=-3 showing the average position of broken beams N=1, N=2, N=3, and N=4.
66
Burst exponents
Z=-1.50(3)
Average over 1000 samples for the burst-size distribution obtained for the injection pressure monitored as a function of the number of breaks. This result agrees very well with other results obtained for catastrophic rupture.
67
Burst exponents
Z=-1.58(2)
Average over 1000 samples for the burst-size distribution obtained for the injection pressure monitored as a function of the number of breaks. Agrees fairly well with other results obtained for catastrophic rupture.
68
Burst exponents
Z=-1.66(2)
Average over 1000 samples for the burst-size distribution obtained for the injection pressure monitored as a function of the number of breaks. No longer agrees with expected Z=-1.50 result for catastrophic rupture.
69
Burst exponents
Z=-1.68(2)
Average over 1000 samples for the burst-size distribution obtained for the injection pressure monitored as a function of the number of breaks. Agrees even less with expected Z=-1.50 result for catastrophic rupture.
70
Burst exponents
If we concentrate on those events that occur closer to the catastrophic point then we see that the result improves. In fact, the improvement gets better the closer we are to the catastrophic failure point.
71
Burst exponents
If we concentrate on those events that occur closer to the catastrophic point then we see that the result improves. In fact, the improvement gets better the closer we are to the catastrophic failure point.
72
Burst exponents
In the case of D>0 disorder the burst exponent converges on the value D=-2 as the disorder magnitude increases.
73
Burst exponents
74