Conference PaperPDF Available

Simulating spalling with a flat-jointed material

Authors:
Applied Numerical Modeling in Geomechanics – 2020 – Billaux, Hazzard, Nelson & Schöpfer (eds.) Paper: 03-01
©2020 Itasca International, Inc., Minneapolis, ISBN 978-0-9767577-5-7
1 INTRODUCTION
The Swedish Nuclear Fuel and Waste Management Company (SKB) is developing a methodology for
long-term storage of nuclear spent fuel in an underground repository in compact crystalline rock (SKB
2011). The repository design calls for the placement of the spent nuclear fuel in copper canisters in verti-
cal 1.8 meter diameter boreholes of 8 meter length. It is likely that the stress redistribution around the ex-
cavations, along with the thermal load from the decaying spent nuclear fuel, will induce stress at the
borehole walls that is greater than the rock mass strength, and thereby produce an excavation damaged
zone (EDZ) in which hydromechanical and geochemical modifications may induce significant changes in
flow and transport properties. The mechanical characteristics of this EDZ must be understood to evaluate
its impact on repository performance.
Our understanding of the damage process can be enhanced by performing in-situ testing. The Äspö Pillar
Stability Experiment (Andersson & Martin 2009, Andersson et al. 2009) is one such in-situ test. The ex-
periment was performed in Äspö diorite and consisted of drilling two vertical 1.75 meter diameter bore-
holes of approximately 6 meter length separated by a one meter pillar. The pillar was loaded by a combi-
nation of excavation-induced stresses and heating of the surrounding rock. In this way, the tangential
stress on the pillar boundary could be controlled in a very precise manner.
In competent crystalline rock, the EDZ takes the form of spalling on the excavation boundaries, whereby
a localized damage region forms at the location of the maximum tangential stress. Spalling begins with
small rock chips forming on the boundaries of the excavation. In the APSE experiment, the chips ranged
in size from a finger nail to approximately 0.1 m2. Larger chips formed beneath the small chips. The
chips are tangential to the hole wall and are believed to have formed in tension. The spalling region
evolves into a v-shaped notch as the tangential stress increases with crushing and/or shearing at the notch
tip, suggesting that the chip-forming process is suppressed at the notch tip. In the APSE experiment, rock
mass yielding occurred when the tangential stress on the hole boundary reached 59% of the unconfined-
compressive strength of the intact rock. Changes in the tangential stress of approximately one MPa be-
yond this threshold expanded the extent of rock mass yielding, indicating that a distinct stress level
threshold needs to be reached before the rock yields.
Our understanding of the damage process can also be enhanced by numerical modeling. This paper sum-
marizes the results of a study to evaluate the ability of the PFC (Itasca 2018) flat-jointed material model
to simulate spalling (Potyondy 2019). Both 2D and 3D flat-jointed Äspö diorite materials were created,
and their response during direct-tension and compression tests was studied. The material behavior is de-
scribed in Section 2. Borehole models of a cylindrical hole in an infinite medium were created and used to
approximate the conditions in the APSE experiment. The behavior of the borehole models is described in
Section 3. Conclusions of this study are provided in Section 4.
Simulating spalling with a flat-jointed material
David O. Potyondy1 & Diego Mas Ivars2,3
1Itasca Consulting Group, Inc., Minneapolis, MN, USA
2Swedish Nuclear Fuel and Waste Management Co (SKB), Solna, Sweden
3Division of Soil and Rock Mechanics, Royal Institute of Technology (KTH), Stockholm, Sweden
2 FLAT-JOINTED MATERIALS
The flat-joint contact model provides the macroscopic behavior of a finite-size, linear elastic, and either
bonded or frictional interface that may sustain partial damage. A flat-jointed material mimics the micro-
structure of angular, interlocked grains. We refer to the balls of a flat-jointed material as faced grains,
each of which is depicted as a spherical core and a set of skirted faces (see Fig. 1a). The flat joint model
formulation is given in Potyondy (2018).
Both 2D and 3D flat-jointed materials to represent Äspö diorite were created and subjected to direct ten-
sion and compression tests. These materials match the Young’s modulus, direct-tension strength, uncon-
fined-compressive strength, and peak strength at 7-MPa confinement. The crack-initiation stress is too
low, the crack-damage stress is too high, and the Poisson’s ratio is too low. No attempt was made to
match the crack-initiation and crack-damage stresses. Attempts to increase the Poisson’s ratio were un-
successful. The material behavior during these tests is like the brittle behavior of compact rock (Potyondy
2015), with the exception that transgranular cracking occurring within and across grains during the com-
pression tests is absent. In compact rock under near-zero confinement, axial splitting occurs, and with in-
creased confinement, shear fractures form. These behaviors are exhibited by the 2D flat-jointed material
but appear to be absent in the 3D flat-jointed material. Although there is extensive dilation after the crack-
initiation point in the 3D flat-jointed material, localizations associated with axial splitting and shear-
fracture formation are absent; instead, the deformation field remains homogeneous.
Microstructural validity describes whether the grain faces overlap one another; a flat-jointed material has
a valid microstructure if and only if the faces of each grain can be connected to the grain center with no
overlap. A valid microstructure is physically realizable i.e., a physical replica of such a material could
be constructed. An invalid microstructure may produce useful behavior, and as such, its use can be justi-
fied. For the present modeling effort, microstructurally valid and invalid instances of the flat-jointed ma-
terial that match the Äspö diorite properties listed above were created (see Fig. 1b). The microstructurally
valid 2D material was selected for the borehole models because it is physically realizable. The micro-
structurally invalid 3D material was selected for the borehole models because the microstructurally valid
3D material was deemed to be unrealistic, having a very small amount of cement with excessively long
cement bridges that overlap across bonded regions and microproperties approximately 3.5 times larger
than their corresponding macroproperties to compensate for the fact that the interfaces do not fully cover
the ball surfaces. This suggests that the 3D flat-jointed material may not be well-suited to model a com-
pact rock like Äspö diorite and may be better suited to model a more porous rock like a sandstone. The
PFC3D bonded-block model that represents the rock as a bonded collection of rigid polyhedral blocks
with an initial porosity of zero may be better suited to model compact rock.
The flat-jointed material exhibits a size effect with modulus and peak compressive strengths increasing
with increasing specimen resolution (number of grains across the specimen width) until a representative
volume is reached. There is no size effect for the tensile strength and Poisson’s ratio. The representative
volumes for the 2D and 3D materials correspond with specimen resolutions of approximately 25 and 50,
respectively. For the 3D material, the values reach approximately 92% of their asymptotic values at a res-
olution of 20. The specimen resolution is the controlling parameter i.e., the same response is obtained
by either varying the specimen size while keeping the average grain diameter constant or varying the av-
erage grain diameter while keeping the specimen size constant.
3 BOREHOLE MODELS
Both two- and three-dimensional borehole models of a cylindrical hole in an infinite medium, have been
developed. The two-dimensional models approximate the conditions for the Kirsch solution (long excava-
tion with a circular cross-section in a medium subject to biaxial stress), whereas the three-dimensional
models are quasi 3D, meaning that plane-strain conditions are enforced on a plane normal to the hole axis
and the model thickness in this direction is small relative to the hole diameter (with approximately four
grains through the thickness). These borehole models include PFC models (both 2D and 3D) and a cou-
pled FLAC3D-PFC3D model. The coupled model allows greater hole resolution (number of grains across
the hole diameter) to be obtained in the near-borehole region.
Figure 1. (a) Flat-jointed material; and (b) microstructural characteristics of the four materials (left: microstructural-
ly valid, right: microstructurally invalid) that match the properties of Äspö diorite.
The borehole models approximate the conditions in the Äspö Pillar Stability Experiment. After excava-
tion of the hole, the maximum tangential stress reaches 165 MPa according to the Kirsch solution, but on-
ly reaches 128 MPa according to a stress analysis accounting for the drift and both holes. A set of coupled
borehole models is constructed for which the material grain size is varied to produce models with a range
of hole resolutions. The displacement field is compatible across the coupling interface (see Fig. 2a), and
when the models are run elastically, the excavation-induced displacements provide a good match to the
Kirsch solution for all hole resolutions. The damage increases with increasing hole resolution, with grad-
ual delineation of a spalling zone and two damage lobes (see Fig. 2b). These damage features become
more clearly delineated as grain size is reduced from 55 to 9 mm. This is approaching the grain size of
Äspö diorite (which ranges from 0.1 to 5 mm). It is expected that the model response will best match the
rock behavior when the average grain size of the model is equal to that of the rock; therefore, the general
damage characteristics are studied for the model with the 9-mm grain size. This model has 718,000 balls
with a total run time of 48 hours on a PC with an Intel Xeon CPU at 3.4 GHz.
Figure 2. (a) Excavation-induced displacement of the coupled model with 9-mm grain size showing displacement-
magnitude contours of the balls and zones. (b) Damage summary overlaid on damage after excavation. The crack
magnification is chosen to draw each crack as a disk with a 36-mm diameter. When the cracks are magnified in this
way, the crack density cannot be visually assessed, appearing larger than it is because the neighboring cracks over-
lap one another.
Damage consists of a spalling zone and two damage lobes. The spalling depth is 8 grains over a 100-
degree sector and reduces to sparse single grains over the remaining perimeter. The damage lobes consist
of a crack swarm located at 40 degrees from the horizontal and extending by the distance of the hole radi-
us into the rock. The crack density and dilation are greater in the spalling zone than in the damage lobes,
and the swarm cracks are aligned with the compressive force chains. The spalling zone corresponds with
an early stage of rock mass yielding before a well-defined v-shaped notch has formed. The cracks in the
spalling zone are like the rock chips that formed in the APSE experiment: formed in tension, aligned par-
allel to the surface, and dilated. Discrete chips have not formed in the model; such chips might form if the
model grain size was reduced to equal the grain size of Äspö diorite. It is likely that the damage lobes are
formed where the stress is near to or greater than the crack-initiation stress. The crack-initiation stress of
the model material is less than that of the rock (60 versus 90 MPa, respectively); thus, these lobes may be
present only in the model and not in the APSE experiment. They were not detected by the AE monitoring
of the experiment; however, damage in the lobe regions may be occurring as individual (not clustered)
grain-scale cracks that have not been detected by conventional acoustic emission monitoring. The damage
lobes are also present in a 2D borehole model of the APSE experiment; however, the damage lobes are
not well formed until the APSE load has been scaled up by a factor of two.
Two variants of the base material that match the Äspö diorite properties listed above were created and
used in the borehole modeling. In the first material, 25% of the fully bonded interfaces were replaced with
slits to increase the Poisson’s ratio. In the second material, a distribution of microstrengths was used. The
overall damage characteristics (spalling zone and two damage lobes) were only minimally affected by
these material modifications.
4 CONCLUSIONS
The PFC3D flat-jointed material can be used to model spalling. By using a coupled FLAC3D-PFC3D
model, a 9-mm grain size for a model that approximates the 1.75 meter diameter APSE experiment could
be run in a reasonable amount of time (48 hours). This model produced a spalling zone and two damage
lobes. The spalling zone corresponds with an early stage of rock mass yielding before a notch has formed
with tensile cracks aligned parallel to the hole surface and dilated. Work is planned to model the URL
Mine-by Experiment (Martin 1997) using PFC3D flat-jointed, soft-bonded and bonded-block models to
provide a quantitative assessment of PFC3D spall-modeling capability.
REFERENCES
Andersson, J.C. & Martin, C.D. 2009. The Äspö Pillar Stability Experiment: Part I Experiment Design. Int. J.
Rock Mech. & Min. Sci., 46(5), 865878.
Andersson, J.C., Martin, C.D. & Stille, H. 2009. The Äspö Pillar Stability Experiment: Part II Rock Mass Re-
sponse to Coupled Excavation-induced and Thermal-induced Stresses. Int. J. Rock Mech. & Min. Sci., 46(5),
879895.
Itasca Consulting Group, Inc. 2018. PFC3D Particle Flow Code in 3 Dimensions, Ver. 6.0 User’s Manual. Itasca:
Minneapolis.
Martin, C.D. 1997. The Effect of Cohesion Loss and Stress Path on Brittle Rock Strength. Canadian Geotechnical
Journal, 34, 698725.
Potyondy, D. 2019. Micromechanical Spalling Model. Itasca Consulting Group, Inc., Report to Svensk Kärnbräns-
lehantering AB (SKB), Stockholm, Sweden, 2-5732-01:19R16, April 19, 2019.
Potyondy, D.O. 2015. The Bonded-Particle Model as a Tool for Rock Mechanics Research and Application: Current
Trends and Future Directions. Geosystem Engineering, 18(1), 128.
Potyondy, D.O. 2018. A Flat-Jointed Bonded-Particle Model for Rock. paper ARMA 18-1208 in Proceedings of
52nd U.S. Rock Mechanics/Geomechanics Symposium, Seattle, USA, 1720 June 2018.
SKB. 2011. Long-term safety for the final repository for spent nuclear fuel at Forsmark. Svensk Kärnbränslehanter-
ing AB (SKB), Stockholm, Sweden, TR-11-01, March 2011.
... The particle flow code (PFC), which is based on the discrete element method, has been widely used to model failure behavior in rock engineering [35,36]. Compared with other numerical software, the complex fracture criterion is not required, and the crack evolution of rock materials can be observed directly in PFC [37,38]. ...
Article
Full-text available
In deep rock engineering, caverns are often disturbed by engineering loads from different directions. To investigate the dynamic response of deep U-shaped caverns under different incident orientations, a theoretical solution of the dynamic stress concentration factor along the cavern boundary was derived based on the wave function expansion and conformal mapping method, and the failure characteristics around the cavern were further investigated by PFC2D (Particle Flow Code in two dimensions). As the incident orientation increases from 0° to 90°, the dynamic compressive stress concentration area transforms from both the roof and the floor to the sidewalls, and the peak dynamic stress concentration factor of the roof decreases from 2.98 to −0.20. The failure of the floor converts from dynamic compression shear failure to dynamic tensile failure. Compared to a stress wave incident from the curved boundary, a stress wave incident from the flat boundary causes severer damage. When the stress wave is incident from the sidewall, the cavern with a larger height-to-width (h/w) ratio exhibits severer damage. Conversely, the cavern with a smaller h/w ratio tends to fail as the stress wave is incident from the floor. This paper provides a basic understanding of dynamic responses of the deep U-shaped cavern.
Article
Full-text available
The deviation angle of wellbores has a significant effect on sand production. Also, the direction of wellbores according to the field stresses is another effective parameter in increasing (or decreasing) the sand production rate. One of the limitations of modeling in the study of the sand production process is to provide a comprehensive three-dimensional model that can simulate all the real conditions of the reservoir, including the situations of the wells and the cavities. In this study, the variations in the sand production rate according to different wellbore inclinations and azimuths were investigated in one of the sandstone reservoirs of the Asmari formation located in the southwest of Iran. For this purpose, a finite difference model (FDM) was developed. Due to the complexity of the geometry of the reservoir, wellbore, and perforations in different situations and the inability of finite difference meshing in modeling such a structure, finite element meshing (FEM) was used. A discrete element model (DEM) was coupled with the finite difference model to simulate the sandstone reservoir. Fluid flow was added to the model using computational fluid dynamics (CFD). Using this FDM-DEM-FEM-CFD hybrid numerical model, the process of sand production in wellbores with different deviations and azimuths was investigated. The results of numerical models showed that according to the normal stress regime of the reservoir, vertical wells produce less sand in this oilfield than horizontal and deviated wells. Also, the highest amount of sand production is related to deviated wellbores, especially those drilled in the direction of minimum horizontal stress. In addition, in horizontal wells with a flow rate of 2000 bbl/day, the variation in the azimuth does not significantly affect the sand production rate.
Article
Full-text available
Reservoir pressure reduction due to continuous production from oil and gas wells affects the sand production rate. An increase in drawdown pressure and/or a decrease in reservoir pressure increases the sand production rate. Since the problem of sand production is one of the main issues in the Asmari sandstone formation located in one of the oilfields in the southwest of Iran, therefore, in this research, the variations in the sand production rate due to the changes in the reservoir and drawdown pressures were investigated. So, for the first time, a hybrid numerical model of finite difference method (FDM)—discrete element method (DEM)—finite element method (FEM)—computational fluid dynamics (CFD) was developed. This numerical model investigated the increase in the sand production rate due to variations in reservoir pressure with a constant bottom-hole flowing pressure. Then, by performing an extensive sensitivity analysis on different values of reservoir pressure and drawdown pressure, the changes in the sanding rate, the critical drawdown pressure, and the safe drawdown line were determined. The results showed that, if the production flow rate of the well is constant, increasing the drawdown pressure can change the sand production rate only to a certain extent, and more than that, will be produced at a constant rate. Also, adjusting the drawdown pressure within a safe range does not necessarily keep the sand production rate constant at a permissible value for a long time, while by keeping the bottom hole flowing pressure constant within an acceptable range, the sand production rate can be controlled for a longer period.
Article
Full-text available
A detailed modelling approach to represent masonry at the meso-scale is proposed, based on the discrete element method, considering the nonlinear behavior of the joints and the units. The fracture of units is represented by the bonded-block concept, in which a random net-work of potential cracks is created, allowing the progressive development of failure mechanisms. For simplicity, only the 2D case is presented, but the extension to 3D is straightforward. A key component of the proposed model is a framework for a joint or interface constitutive model, in-cluding the post-peak softening range, taking into account the experimental fracture energies. In this model, the softening curves in tension or shear are defined by piecewise linear segments, calibrated to reproduce the most common masonry constitutive models. The essential issues in-volved in the application of bonded-block models to masonry are examined, namely the block shape, either Voronoi polygons or triangles; size; deformability; and the influence of the main constitutive parameters. Uniaxial compression tests are analyzed in detail. The simulation of a well-known experiment of a brick panel under shear shows the good performance of the pro-posed approach. The investigation results demonstrate the model’s capabilities and provide guidelines for its application.
Article
Full-text available
Stress-strain curves for brittle rocks show three characteristic stress levels: crack initiation, long-term strength, and peak strength. Damage-controlled testing at low confining stresses has shown that the long-term and peak strengths are sensitive to the amount of induced damage, i.e., the greater the amount of damage, the lower the long-term and peak strengths. These tests also showed that the brittle-failure process is characterized by a loss of cohesion as friction is mobilized. Excavation of a circular test tunnel in massive brittle rock resulted in failure around the tunnel. The back-calculated strength for the failed rock around the tunnel is approximately one-half of that measured in laboratory tests. Crack-induced damage of Lac du Bonnet granite, both in the laboratory and in situ, begins when the load exceeds approximately one-third of the unconfined compressive strength. However, the stress level associated with failure is a function of loading path. In the laboratory, where the loading path monotonically increases, the ultimate strength of an unconfined sample is 225 MPa. Numerical studies suggest that in situ the loading path around the tunnel is more complex, involving stress increase and decrease and stress rotation. For this loading path, failure initiates at a stress between 100 and 120 MPa. Conventional frictional failure criteria did not adequately predict the extent of brittle failure measured around the circular tunnel. The results from the damage-controlled laboratory tests and the microseismic monitoring carried out during tunnel construction indicate a constant-deviatoric-stress criterion is a reliable indicator for predicting the onset of damage. This criterion was also found to give a reasonable prediction for the maximum depth of failure around the test tunnel. The fundamental assumption in the constant-deviatoric-stress criterion is that at low confining stresses, such as those which occur around underground openings, the brittle-failure process is dominated by cohesion loss.
Article
We generalize our view of a bonded-particle model (BPM) to consist of a base material (that is a packed assembly of rigid grains joined by deformable and breakable cement at grain–grain contacts) to which larger-scale joints can be added and whose mechanical behavior is simulated by the distinct-element method using the two- and three-dimensional discontinuum programs PFC2D and PFC3D. The micromechanical processes that control brittle fracture, and thus, should inform any micromechanical theory or model, are summarized. The rich variety of microstructural models that can be produced by the bonded-particle modeling methodology are described and classified with respect to their microstructural and larger-scale features. These models provide a wide range of rock behaviors that encompass both compact and porous rock at both an intact and rock-mass scale, and examples are provided of how BPMs are being used to model rock at these scales. The examples include an intact anisotropic material that may swell and contract in response to changes in saturation, the behavior of two alternative BPMs that can match both the uniaxial and tensile strengths of compact rock and the embedding of an intact BPM within a larger continuum model to study fracturing around a gold-mine stope in quartzite.
Article
A 1-m-thick pillar was subject to coupled excavation- and thermal-induced stresses to induce brittle rock mass yielding. The yielding strength of the heterogeneous and fractured rock mass consisting of Äspö diorite was evaluated at eighteen discrete locations using data from the displacement, acoustic emission, and thermal monitoring systems. The average rock mass yielding strength was determined to be 0.59 of the uniaxial compressive strength.The onset of dilation in uniaxial laboratory tests, determined from strain gauge data, was found to occur at approximately 0.45 of the uniaxial compressive strength. It was shown that that the onset of acoustic emission events in situ also occurred when the tangential stress exceeded 0.43 of the uniaxial compressive strength. For sites with absence of in situ data it is recommended that this lower-bound value determined from laboratory data may be used for assessing the in situ rock mass yielding strength.Visual observation and displacement monitoring showed that extent of rock mass yielding is sensitive to small changes in the tangential stress magnitudes. It was determined using three-dimensional modelling that changes in the tangential stress magnitude of approximately 1MPa was sufficient to cause yielding of the pillar to propagate in what appeared to be intact rock. Observations suggest that without this small stress change yielding of the rock mass would not occur. In other words, there appeared to be a well defined boundary, and if the stresses reached this boundary yielding was observed. However, if stresses were only slightly below this boundary yielding or time-dependant processes were not observed over the monitoring period used in the experiment.
Article
The Äspö pillar stability experiment was carried out to examine the failure process in a heterogeneous and fractured rock mass when subjected to coupled excavation-induced and thermal-induced stresses. The rock pillar was created by the excavation of two adjacent large-diameter boreholes. The pillar was loaded by a combination of excavation-induced stresses and heating of the surrounding rock by a rectangular heater pattern. The experiment was designed using the observation design method. The characterzation of the experiment volume showed that the experiment was located in a fractured water-bearing rock mass that was considered typical for the Äspö Hard Rock Laboratory (Äspö HRL) and for the Fenno-Scandinavian shield. Scoping calculations using two- and three-dimensional elastic stress analyses were carried out to reduce the uncertainty for the far-field in-situ stresses, establish the geometry for the access tunnel that would provide a elevated uniform stresses in the floor of the tunnel, and determine the optimum width of the pillar. It was concluded, based on the rock mass characteristics, that a 1 m-wide pillar formed by two 1.75 m diameter boreholes would meet the design objectives. Thermal modelling showed that thermally induced stresses in the pillar were adequate to elevate the pillar stresses above the stress magnitude required to initiate failure.Acoustic emission, displacement, and thermal monitoring systems were installed according to the experiment design without problems. No sensors were lost over the three month duration. After the experiment was completed, a laser scan of the pillar revealed the extent of the damaged pillar. The experiment commenced in January 2002 and was successfully completed in 2006.
Micromechanical Spalling Model
  • D Potyondy
Potyondy, D. 2019. Micromechanical Spalling Model. Itasca Consulting Group, Inc., Report to Svensk Kärnbränslehantering AB (SKB), Stockholm, Sweden, 2-5732-01:19R16, April 19, 2019.
Long-term safety for the final repository for spent nuclear fuel at Forsmark. Svensk Kärnbränslehantering AB (SKB)
  • Skb
SKB. 2011. Long-term safety for the final repository for spent nuclear fuel at Forsmark. Svensk Kärnbränslehantering AB (SKB), Stockholm, Sweden, TR-11-01, March 2011.