Sei sulla pagina 1di 15

URTeC: 420

Fault Risk Assessment Using Quantitative Structural Geology


Techniques
N. Eichelberger1, W. B. Hawkins*2; 1. StructureSolver LLC, 2. GeoFlite Solutions LLC.
Copyright 2019, Unconventional Resources Technology Conference (URTeC) DOI 10.15530/urtec-2019-420

This paper was prepared for presentation at the Unconventional Resources Technology Conference held in Denver, Colorado, USA,
22-24 July 2019.

The URTeC Technical Program Committee accepted this presentation on the basis of information contained in an abstract
submitted by the author(s). The contents of this paper have not been reviewed by URTeC and URTeC does not warrant the
accuracy, reliability, or timeliness of any information herein. All information is the responsibility of, and, is subject to corrections by
the author(s). Any person or entity that relies on any information obtained from this paper does so at their own risk. The information
herein does not necessarily reflect any position of URTeC. Any reproduction, distribution, or storage of any part of this paper by
anyone other than the author without the written consent of URTeC is prohibited.

Abstract
Quantitative structural geology techniques can be used in conjunction with seismic data to define fault
locations and geometry to reduce risk when planning horizontal wells. Unconventional plays in the past
decade focus on basins characterized by largely horizontal stratigraphy and minimal faulting. However,
even basins with minimal structure contain small-scale faults that pose unique risks to horizontal drilling.
A fault with 100’ of throw or less can present well-bore stability issues due to natural fracturing and rock
strength contrasts across faults. Seismic data can image these faults but because of their subdued nature,
reliably interpreting their geometry and understanding the associated drilling risks can be difficult.

We present case studies examining seismic sections from the Denver basin where horizontal wells
intercept small-scale normal faults. These faults are visible on the seismic sections where they cut across
the Niobrara at 30 – 45°. Wells that pass above, below, or intersect the faults at high angles are generally
drilled and completed without issue. At deeper levels, wells intersected the same faults at very low angles
(where fault dips < 30°) and experienced drilling and completion issues. The transition from steep fault
dips within the Niobrara to low fault dips at deeper levels is not well resolved in the seismic sections but
clearly represents a drilling hazard that should be considered.

The geometry of the faults can be estimated using structural forward modeling and area-depth-strain
(ADS) analysis. These established structural geology methods quantitatively relate seismically-imaged
fold shape and horizon displacements to the underlying fault geometry. We use these methods to compute
the deep fault geometry from the seismically observed folding of the Niobrara formation. We demonstrate
that the wellbore collapse was related to fault angle and for a shorter lateral distance. This strongly
suggests that the listric bends in these faults is a high-risk zone.
URTeC 420
2
Introduction
In this paper, we demonstrate how quantitative structural geology techniques can be used in conjunction
with seismic data to define fault locations and geometry to reduce risk when planning horizontal wells.

Unconventional plays in the past decade focus on basins characterized by largely horizontal stratigraphy
and minimal faulting. However, even basins with minimal structure contain small-scale faults that pose
unique risks to horizontal drilling. In many cases, the target formations may be on the order of 10s of feet
thick, so faults with 100’ of throw or less can produce significant stratigraphic offsets that result in well-
bore stability issues due to rock strength contrasts across faults and natural fracturing.

Seismic data can image and help define the risks associated with faults prior to drilling. Accurately
assessing the drilling risk presented by a given fault requires a robust interpretation of the fault geometry
and fault throw. However, the structures present in many unconventional plays are extremely subtle and
close to the limit of seismic resolution, making seismic interpretations highly uncertain. For example,
faults in the DJ Basin (Figure 1) have been interpreted qualitatively as either listric (Figure 2) or planar
(Figures 3 and 4). Both interpretations are correct throughout the basin, but the fault geometries are not
always clearly defined in seismic sections.

Using established structural geology techniques, we can quantitatively predict fault geometries and
throws even in situations where the faults are not directly resolved in the seismic image. In this study, we
apply structural forward modeling (Xiao and Suppe, 1992) and area-depth-strain (ADS) analyses (Epard
and Groshong, 1993) to seismic sections along horizontal wells targeting the Niobrara Formation in the
DJ Basin. The studied sections indicate the presence of faults with throws on the order of 5’ – 50’. The
predicted fault geometries and throws from structural modeling and ADS analysis are compared against
seismic and geosteering interpretations. We find that the structural geology methods accurately estimated
fault locations and throws compared to these interpretations. One well in particular encountered drilling
and completion issues where it incepted a listric fault at a low angle. The seismic data for this well clearly
images a half-graben but does not directly image the listric nature of the underlying fault. Structural
modeling and ADS analysis both accurately predicted the fault curvature from the seismically-imaged
geometry of the half-graben. Using these techniques in conjunction with seismic data during the well
planning process would have indicated the presence of a listric fault, identifying it as a high-drilling risk
zone that should be avoided.
URTeC 420
3

Figure 1. Structure contour map on the top of the Niobrara Formation in the Denver Basin. Study wells and stratigraphic column with vertical
drilling depths, from Wattenberg Field. Modified from Sonnenberg (2015).

Figure 2. Diagram depicting the Davis (1985) listric fault model.


URTeC 420
4

Figure 3. Schematic diagram of a multi-layer polygonal fault system. Faults are layer-bound, randomly-oriented, are normal in character, and
create polygonal shapes from traces when looked at in map view (Cartwright, 1994).

Figure 4. A seismic line that displays a series of 45º planar, normal faults that are layer bound. Note that the normal faults bound within the
Niobrara are not interpreted to extend far into the underlying Carlile Shale (Sonnenberg and Underwood, 2013)
URTeC 420
5
Methods
We examined four wells from Wattenberg Field of the Denver basin which intercept normal faults within
the Niobrara Formation and Codell Sandstone Member of the underlying Carlile Shale (Figures 5-13).
These faults are visible on the seismic sections where they cut across the Niobrara at 30º – 45°. Wells that
pass above, below, or intersect the faults at high angles are generally drilled and completed without issue.
At deeper levels, wells intersected the same faults at very low angles (where fault dips < 30°) and
experienced wellbore collapse, rock instability, and completion issues. The transition from steep fault dips
within the Niobrara to low fault dips at deeper levels is not well resolved in the seismic sections but
clearly represents a drilling hazard that should be considered.

To overcome these seismic imaging limitations, the geometry of the faults were estimated using structural
forward models and area-depth-strain analyses. These established structural geology methods
quantitatively relate seismically-imaged fold shape and horizon displacements to compute the underlying
fault geometry. Area-depth-strain analysis is performed by measuring areas of horizons folded in the
hangingwall of a fault relative to the horizon’s footwall depths. For pre-kinematic horizons, plotting the
fold areas as a function of footwall depth generally reveals that hangingwall fold areas linearly decrease
with depth towards the underlying fault. Using a simple linear regression allows us to compute a
predicted fault depth based on the expectation that fold areas will decrease to zero at the depth of the
fault. Therefore, the y-intercept on the area-depth graph (depth intercept) gives the depth of the fault and
the slope of the linear fit is the fault displacement (for details, see Epard and Groshong, 1993). Structural
forward models, sometimes referred to as fault-bend-fold models, are geometric models for computing
the fold geometries that arise as rocks are displaced over curved faults based on basic kinematic
assumptions (such as conservation of area). In this paper we used the inclined shear method from Xiao
and Suppe (1992).

We demonstrate that the wellbore collapse was related to fault geometry.

Wells A-C were drilled primarily in the Codell Sandstone. Well D was drilled predominantly in the
Niobrara B Chalk. The suite of wells documents a range of structural deformation encountered within the
Denver Basin. Small scale faults on the order of <10’ are undetectable on seismic but are often detected
while geosteering (Figure 11).

Results
The structural analysis reveals that potentially risky faults are not planar. Instead these faults transition
from a relatively steep dip in the Niobrara to a flat detachment within and below the Codell. Well D
collapsed where it intercepted the fault bend from high dip in the Niobrara to the flat detachment near the
Codell/Carlile contact. Consequently, well placement should pass through the fault damage zone at a
higher angle and for a shorter lateral distance. This strongly suggests that the listric bends in these faults
is a high-risk zone.
Another finding when restoring the grabens and half grabens was that the graben in Well C models with
an anomalously deep detachment when using the depth to detachment method (Figure 8b).

Well A Results
URTeC 420
6
The seismic line for Well B features two subtle structures, a full graben and a half graben (Figure 5). The
controlling faults for both are not directly imaged in the seismic section and geosteering interpretations
suggest only minor throw on the faults.

Figure 5. Well A uninterpreted seismic section.

The full graben is ~800’ in length, characterized by a slight symmetrical depression in the seismic
reflectors that is consistently repeated across 500’ in depth. To reproduce the seismic image of the graben
using a structural forward model, two planar fault models are necessary to explain the symmetrical shape
of the depression and the flat dip of the seismic reflections within the depression (Figure 6A). Structural
forward models that match the seismic image estimate no more than 20’ of throw and possible fault dips
between 50º – 60º. The structural model estimates 35’ of extension across the full graben.

Manually interpreting the graben geometry and performing ADS analysis predicts a detachment depth of
~2000’ (Figure 6B). This is slightly deeper than where one might interpret the bottom of the graben based
on the seismic image where it appears to bottom out just above the Codell Sand. Given the relatively
minor nature of the graben, interpreting fault throw and fault dip is highly ambiguous.

The half graben is characterized by an asymmetrical depression that is ~1,800’ wide and consistently
imaged across a 500’ depth interval. Modeling the seismically imaged structural geometry requires a
controlling fault with a curved, listric geometry (Figure 6A). The model indicates that the upper fault
segment dips at 30º – 40º and soles into a flat detachment within the Carlile Shale. The model estimates
fault throws of no more than 25’ – 30’ and total extension of 68’. ADS analysis of a manual interpretation
of the graben also gives an estimated detachment depth within the Carlile Shale and total extension of 80’,
broadly consistent with structural forward model results (Figure 6C).
URTeC 420
7

Figure 6. Well A. A graben and half graben structural models, B and C are graben and half graben ADS models and plots respectively.

Well A Discussion

Both the models, seismic data, and geosteering interpretations indicate that these faults are subtle with
minimal throw. The half graben controlled by the listric fault appears to be a more significant structure as
is has accommodated at least twice the total displacement as the full graben. Geosteering from Well B
did not provide constraints on fault location, but structural modeling indicates that the well intercepted
both structures where the controlling faults had dips greater than 30º. The minor fault throws likely made
it extremely difficult to recognize the presence of the faults from geosteering data. Well B was drilled and
completed without issue.

Well B Results

The seismic line for Well B (Figure 7) only involves a full graben although in this case, the bounding
fault locations are better imaged compared to Well A (Figure 5). The graben at Well B is 1,300’ wide in
length (500’ wider than at Well A) and is seismically imaged across the entire Niobrara Chalk (similar to
URTeC 420
8
Well A). The bounding faults may dip between 50º – 60º as indicated by discontinuities in the seismic
reflections but despite the clear fault breaks in the seismic reflectors, the exact fault shape is uncertain.

Structural forward modeling reproduces the seismically imaged structure using two opposed faults
dipping at ~60º (Figure 8A). The models estimate 40’ – 50’ of throw on each fault and 100’ of total
horizontal extension across the entire structure.

Figure 7. Well B uninterpreted seismic section.

Manual interpretation and ADS analysis of the graben computes a total horizontal extension of ~120’
across the entire graben (Figure 8 B and C). The detachment depth computed from the structural area in
the interpretation is estimated to be in the Graneros Shale. This is substantially deeper than indicated by
the seismic image, where the graben appears to terminate within the Carlile Shale. This could be the result
of interpretation error. To determine if this was the case, we used the fault dips and horizon correlations
suggested by the structural forward model in a separate ADS analysis. The detachment depth computed
from the model-based interpretation indicated a detachment well below the geologically reasonable range
expected for the graben.
URTeC 420
9

Figure 8. Well B. A graben structural models, B and C are graben ADS models and plots.

Well B Discussion

Ultimately, the discrepancy for the graben detachment computed using each method may indicate that the
graben is a polygonal-fault type structure generated by dewatering rather than tectonic deformation. A
URTeC 420
10
key assumption in both structural forward modeling and ADS analyses is that area is maintained during
deformation. Generally, for seismic lines oriented parallel to tectonic transport, area is largely conserved
and applying these methods yields accurate results. In structures settings where area is not conserved,
such as those involving a mobile substrate (such as salt or shale), the computed detachments and fault
shapes will not be consistent with seismic images. Polygonal faulting driven by dewatering does not
conserve area (Cartwright, 1994). For the seismically-imaged grabens here, the discrepancies in the
computed detachment depth may indicate that dewatering, rather than tectonics, are responsible for the
development of the grabens. Conversely, the structural models and ADS analyses of the half-grabens
driven by listric faulting are completely consistent with the seismic data. Furthermore, the listric faults are
observed to have greater throw, indicating they have accommodated larger strains than the planar faults
bounding the full grabens. Collectively, we propose this indicates that the listric faults and associated
half-grabens are tectonic in nature while the planar faults and associated full grabens are polygonal fault
systems.

While the computed detachments from modeling and ADS were at odds with the seismic, the clear
imaging of the faults made modeling and interpreting fault geometry and throw very straightforward. For
this line, both the models and ADS analysis indicate the graben had greater throw than the comparable
full graben on Well A (50’ on Well B versus less than 30’ on Well A). The models for the graben in Well
B indicate that the bounding faults are planar with dips between 50º – 60º, similar to the graben in Well
B. As with Well A, Well B intercepted the graben and was drilled without issue. However, well
completion stages in and adjacent to the graben experienced hydraulic fracturing initiation issues.

Well C Results

The seismic line for Well C (Figure 9) is characterized by two half-graben structures which are
stylistically similar to the half-grabens controlled by listric faults in the seismic lines for Wells A (Figure
6) and D (Figure 13). Both half grabens in the Well C seismic section have clearly imaged, offset seismic
reflectors but the actual fault geometry is not resolved. In the hanging wall of both structures, the seismic
reflectors are clearly folded in a roll-over style fault-bend-fold. The trough that defines the half-graben on
Half-Graben 1 (left side of Figure 10) is ~1,500’ in width while Half-Graben 2 (right side of Figure 10) is
nearly 2,500’ in width.

Figure 9. Well C uninterpreted seismic section.


URTeC 420
11
Structural models (Figure 10A) that best match the seismically imaged grabens indicate that they are
controlled by faults with upper segments that dip 25º – 35º as they cut through the Niobrara Formation
and gradually flatten with depth. The model for Half-Graben 1 indicates a detachment within the Codell,
55’ of throw, and 110’ of total displacement while the model for Half-Graben 2 estimates a deeper
detachment in the Carlile Shale, greater throw at 85’, and more total displacement at 220’ (Figure 10A).
The models demonstrate that the controlling faults for both structures are not planar and instead curve into
a flat detachment.

ADS analysis of manual interpretations of the structures are consistent with model results, both in terms
of detachment depths, throws, and total displacement. Importantly, the ADS analyses confirm that both
structures are controlled by faults that merge into flat detachments and that Half-Graben 1 detaches at a
slightly shallower depth than Half-Graben 2 (Figures 10B and 10C))

Figure 10. Well C. A graben structural models, B and C are graben ADS models and plots.

Well C Discussion
URTeC 420
12
The ADS analyses and structural forward models collectively indicate that Half-Graben 2 is more
structurally significant than Half-Graben 1 and has accommodated greater strains. This is significant
because Well C passed through, or slightly below, the detachment for Half-Graben 1 without issue and
significant faulting was not directly observed in the geosteering data (Figure 11B). In contrast, Well C
passed through the lower portion of the fault in Half-Graben 2 where the structural model indicates that
the fault transitions from a 30º dip into the flat detachment. In this area, the geosteering log indicates a
significant fault contact. During completion of the well, packers were not able to be run past this area,
preventing the remaining ~3000’ of the Well D lateral from being completed. Both structural analyses
methods indicated that both faults had lower flat detachments and indicated curving listric faults with flat
detachments, rather than planar faults that terminated at depth

Figure 11. Well C geosteering interpretation. A. Interpreted seismic. B. Geosteering interpretation with gamma ray and delta gamma logs along
bottom.

Well D Results

Well D is drilled primarily through the Niobrara B Chalk with a faulted section of the Niobrara A Marl.
The seismic line for Well E contains a half graben and a small fault approaching the lower limit of
seismic resolution (Figures 12 and 13). As in earlier examples the half graben rollover is clearly imaged
and seismic reflectors show offset, but the lower fault geometry is ambiguous. The trough that defines the
URTeC 420
13
half-graben (Figure 13) is ~1,000’ in width, with 53’ of throw, and 85’ of total displacement.

Figure 12. Well D uninterpreted seismic section.

The structural model for the seismically imaged half graben (Figure 13) indicates that the controlling fault
steepens with depth. The upper fault segment has a dip of 40º as it cuts through Pierre Shale and Upper
Niobrara and then steepens to ~50º in the Middle to Lower Niobrara before it gradually flattens with
depth. The model for Half-Graben indicates a detachment within the lower Codell/Carlile Shale. The
small fault (right side of Figure 13) has a dip of 58º, a throw of 18’, and 21’ of displacement.

Figure 13. Well D half graben structural model.

Well D Discussion

The structural forward model shows that the even with greater throw a steeper fault segment within the
Niobrara has a small drilling and completion risk associated with it. Because the well encountered the
faults at a high dip angle and the rock juxtaposed on either side of the faults has relatively similar
mechanical properties no drilling or completions issues were had with this well. The half graben seen in
Well D detaches above the Fort Hays Limestone within the lower Niobrara C Marl or the shale rich
Niobrara D Chalk.
URTeC 420
14
Conclusions

Area-depth-strain analysis and structural forward modeling can both be used to define relatively precise
fault geometries from seismically-imaged structural geometries. In this case study from the DJ-Basin, we
were able to quickly define fault geometries, fault throws, total displacement, and differentiate between
structures controlled by planar faults (e.g. the full grabens) and those controlled by curving listric faults
with flat detachments (e.g. the half grabens).

Structural models and ADS results for the full-grabens consistently indicated relatively shallow
detachments, minor fault throws, and low total displacements. Collectively, this indicates the full grabens
accommodated relatively minor amounts of strain. In addition, the computed detachments for the full-
grabens were inconsistent with the seismic data, suggesting that they may be generated by polygonal
faulting (and dewatering) rather than tectonic deformation. Where wells intercepted these structures, there
was little indication for deformation and no completion or development issues were experienced.

The structural geologic analysis of the half-graben structures indicated that they have the potential to be
more structurally significant than the full grabens. Both the structural forward models and ADS analyses
of the half-grabens indicate that they are controlled by curving listric faults with flat detachments. For the
Well C seismic line, the half grabens have far greater fault throws, and total displacements than the full-
grabens from any other line. This indicates that half-grabens in Well D have accommodated higher
strains. For the more significant half-graben in Well D, the drilling risk proved to be high as the wellbore
collapsed where it encountered the listric fault.

Implications for Drilling Risk

While the drilling risk for the full grabens was low, the drilling experience for the half-grabens is more
complex. Wells B and D both intercepted the listric faults controlling half-graben structures. Well B
intercepted a subtle half graben with minor model-estimated throw (< 20’) and total displacement (30’).
This a minor level of deformation, comparable to the throws and displacements estimated for the full-
grabens. As mentioned above, Well B experienced no drilling or completion issues.

Well D intercepted a more significant listric fault below a half graben with model estimates of 50’ of fault
throw and 110’ of total displacement (Half-Graben 1, Figure 10). There were no issues with Well D in
this area because it passed within or just below the flat detachment of the listric fault.

The highest drilling risk was encountered where Well D crossed a high throw (80’) listric fault at a
location where models indicate the fault curves from steeply dipping to the flat detachment. This is the
region where the well bore collapsed during completion.

Retrospectively, the drilling experience seems to corroborate the structural geologic analysis results:
grabens and half-grabens with relatively low throw and total displacement are minor and are not a
significant drilling risk. The highest risk is posed by listric faults with larger displacements. By using
seismic data in conjunction with area-depth-strain analysis and structural forward modeling, the fault
geometry and displacements can be estimated in advance to determine the risk posed by a given structure.
Mapping out fault geometry and detachment locations can then be used to avoid high-risk fault zones.
URTeC 420
15
Acknowledgments
The authors wish to thank Noble Energy Inc. for releasing the seismic and geosteering data used in this
study.

References
Cartwright, J. A. 1994. Episodic Basin-wide Hydrofracturing of Overpressured Early Cenozoic Mud-rock
Sequences in the North Sea Basin. Marine and Petroleum Geology 11(5): 587-607.

Davis, T. L. 1985. Seismic Evidence of Tectonic Influence on Development of Cretaceous Normal Faults,
Boulder-Wattenberg-Greeley Area, Denver Basin, Colorado. The Mountain Geologist 22(2):47-54.

Epard, J.-L., and R. H. Groshong Jr., 1993. Excess area and depth to detachment. AAPG Bulletin,
77(8):1291–1302.

Sonnenberg, S. A. 2015. Geologic Factors Controlling Production in the Codell Sandstone, Wattenberg
Field, Colorado. Presented at the AAPG Annual Convention and Exhibition, Denver, Colorado, 31 May-3
June. AAPG Datapages/Search and Discovery Article #90216

Sonnenberg, S. A. and D. F. Underwood, 2013, Polygonal fault systems—A new structural style for the
Niobrara Formation and Pierre Shale, Denver Basin, Colorado: The Rocky Mountain Geologist 50 (4):
127–142.

Xiao, H.-B., and J. Suppe, 1992. Origin of rollover: AAPG Bulletin, 76(4): 509–529.

Potrebbero piacerti anche