Вы находитесь на странице: 1из 16

SPE-191634-MS

Real-Time 3D Computer Vision Shape Analysis of Cuttings and Cavings

Runqi Han, Pradeepkumar Ashok, Mitchell Pryor, and Eric van Oort, The University of Texas at Austin

Copyright 2018, Society of Petroleum Engineers

This paper was prepared for presentation at the 2018 SPE Annual Technical Conference and Exhibition held in Dallas, Texas, 24-26 September 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
Excessive cuttings and cavings accumulation due to poor hole cleaning and borehole instability can cause
costly stuck pipe incidents. Currently, there is no surface instrument to monitor cuttings volume and detect
cavings in real-time. An automated 3D real-time computer vision monitoring system can quantify cuttings
return volume, detect cavings presence, and analyze cavings shape. This makes pro-active prevention and
mitigation of non-productive time (NPT) caused by poor hole cleaning and wellbore instability possible.
In this paper, we present a real-time computer vision system to measure cuttings properties and detect
cavings. The proposed design consists of a 2D high-resolution camera and a 3D profile laser scanner, which
collect point cloud/depth data of cuttings/cavings after passing the shale shaker. We apply cutting-edge
computer vision algorithms and feature recognition techniques to quantify cuttings volume, detect cavings,
and characterize cavings shape. The angularity, flatness, and other geometrical features of cuttings/cavings
can be determined from the point cloud 3D data.
A prototype computer vision system was constructed and tested in the lab and test yard to evaluate the
system capability to measure cuttings/cavings properties. In a controlled laboratory environment, a sensing
algorithm was designed and tested in the presence of drilling fluid. To improve measurement accuracy, both
artificial and field cavings were used to simulate realistic scenarios and train a data pool. The system was
then validated in a test yard shale shaker testing facility. The accuracy, repeatability, and robustness of the
sensors were evaluated against external lighting variances, dust, humidity, etc. The proposed automated
cuttings/cavings monitoring system can identify cavings and analyze shape characteristics. By diagnosing
potential hazards, the system warns the driller on adverse wellbore conditions and the likelihood of stuck
pipe events. This paper proposes and demonstrates a novel 3D depth-sensing system to measure cuttings
volume, identify abnormal cavings, and analyze shape. This state-of-art, non-intrusive system evaluates
cuttings/cavings quantitatively and delivers algorithms that automate downhole condition monitoring to
reduce drilling-related NPT in the field.

Introduction and Background


During well construction, cuttings and cavings monitoring while drilling is important for examining hole
cleaning efficiency and diagnosing wellbore instability. Inadequate hole cleaning often leads to costly
drilling problems, such as pipe sticking, premature bit wear, slow drilling rate, formation fracturing, high
2 SPE-191634-MS

torque and drag, mud losses etc. (Azar and Samuel 2007). Wellbore instability issue is one of the main
concerns causing NPT in drilling operation. Globally, it costs oil companies approximately over US$ 6
billion each year (Yongfeng Kang 2009). The most common practice to analyze drilling cuttings and cavings
is that a mud logger manually collects and examines cuttings samples continuously. The mud logger's duties
typically include collecting cuttings samples, visual inspection of cavings, checking drilling mud rheology,
and moving back and forth between the mud logging unit and the shale shakers. Not only is his/her health
and safety jeopardized both by inhaling vapors and by direct mud to skin contact (IPIECA 2009), but also
by ascending and descending stairs with an associated risk of injury. Furthermore, this type of monitoring is
only real-time when the mud logger is actually at the shale shakers, which is only infrequently. Therefore,
automating the cuttings and cavings monitoring process is expected to benefit both drilling operations as
well as personnel health and safety improvement.
This paper builds on earlier work by Han et al., 2017 by incorporating real-time cavings detection and
shape classification on reconstructed 3D profiles of cuttings and cavings. Algorithms were trained and
evaluated in the laboratory environment. A combination of cuttings and different types of cavings were
tested and examined based on their geometrical properties. Previously identified geometrical features were
then used to train a classifier to determine the shape of cavings: splintery, tabular, or angular. A preliminary
yard trial was also conducted to assess the system accurancy in an outdoor environment.

Background
In general, two different approaches have been applied in the industry to automate hole cleaning monitoring:
cuttings flow meters (CFM) and cuttings transport models. A cuttings flow meter was first introduced by
Naegel et al., 1998. The basic principle was to measure the weight of cuttings coming off the shale shaker.
A mud-effect correction factor called the Equivalent Dry Cuttings Ratio (EDCR) was applied to correct the
errors caused by the coating of mud on the cuttings. This allows for the volume of equivalent dry cuttings to
be calculated. A real-time cuttings monitoring system using a similar concept design (Ferranod and Hbaieb
2015) was also commercialized. However, drilling fluid not only coats the surface of the cuttings, but is also
absorbed by the porous cuttings. The amount of fluid invasion in the cuttings depends on mud rheology,
lithology of the cuttings, temperature, and interaction time. Various factors may affect the EDCR accuracy.
In addition, the presence of cavings cannot be detected using this system.
Another approach is to simulate cuttings transport via fluid modeling along the wellbore. A cuttings
transport model can be applied to predict cuttings return volume, cuttings bed accumulation, etc. Research
on cuttings transport has been conducted for over 70 years since Pigott's early analysis on factors affecting
the mud's cutting carrying and releasing properties (Pigott 1941). Overall, due to the complexity of the
drilling fluid flow behavior, early studies focused primarily on experimentally analyzing the effects of
various parameters like drilling mud rheology, pipe inclination, pipe rotation, flow rate etc. Various cuttings
slip velocity models were formulated based on cuttings particle size, cuttings density, drilling fluid density,
particle irregular shapes, Reynold's number etc. (see e.g. Chien 1972, Zeidler 1972, Moore 1974, Cranford
1992, C.J. Hopkins 1995, Rubiandini 1999, and Baldino, et al. 2015). Experimentally, large-scale cuttings
transport experimental flow loops were built and various models and parameters were tested to better
understand the multi-phase behavior of cuttings transport in the wellbores (see e.g. Miska et al. 2004). As
interest in deviated and horizontal wells increased and underbalanced drilling became more important as
an essential practice, the interest expanded to mechanistic modeling of cuttings transport in the presence of
multi-phase flow. More comprehensive transport models were studied and developed based on modeling
of cuttings layers, cuttings bed accumulation, varying behavior of various phases and other mechanisms
(see e.g. Wilson 1974, Gavignet and Sobey 1989, Larsen 1990, Martins and Santana 1992, Jalukar 1993,
Clark and Bickham 1994, Nguyen and Rahman 1996, Ramadan 2001, Li et al. 2007, Salazar-Mendoza et
al. 2008 etc.).
SPE-191634-MS 3

Recent studies have focused on developing multi-phase transient cuttings transport model while
incorporating mechanistic models developed earlier. In 2013, Cayeux et al., proposed a comprehensive
real-time transient cuttings transport analysis model. The transient drilling hydraulics model was governed
by mass-conservation, momentum balance, and energy conservation of solid and fluid phases. Cuttings
suspension, transport, settling and bed accumulation/erosion were also taken into consideration and
analyzed. The model was validated on two field cases and reflected hole cleaning problems. However, due to
the lack of automated surface monitoring equipment, it proved to be a challenge to compare and analyze the
downhole simulation and surface observations. No work has been done since comparing cuttings transport
simulation results with real-time surface measurement of cuttings’ return volume and size distribution in the
field. In addition, cavings which result from wellbore instability at present cannot be accurately simulated
through cuttings transport modeling.
Currently, the drilling industry relies on a mud logger to monitor the shale shaker 24/7 while attempting
to spot and differentiate cavings from drilling cuttings. This is a tedious and labor-intensive process and is
neither thorough nor structured. As a result, cavings are not properly and timely identified and analyzed.
Cavings are the first and foremost indicator of wellbore deterioration. Immediate and correct interpretation
of cavings can help save millions of dollars through appropriate remedial actions (Aldred et al. 1993). To
keep track of the health of wellbore, it is crucial to monitor and analyze the size, shape, appearance and
relative percentage of cavings compared to the total load of drilling cuttings from shale shaker. Monitoring
the rate at which cavings are produced and analyzing the morphology can help identify wellbore failure
mechanism (Osisanya 2012). The current method to detect cavings relies entirely on human visual inspection
and manual sampling of shale shaker screens. A study was conducted by Kumar et al. (2012) wherein they
visually distinguished cavings and cuttings and characterized cutting shapes into three categories: splintery,
angular, and tabular (Figure 1). After analyzing the causes and failure mechanisms based on cavings type
(Table 1), they were able to conduct remedial actions, such as changing drilling mud weight to provide
mechanical wellbore support, to migrate wellbore instability issues. Therefore, it is crucial to detect and
analyze the first cavings returning to surface to address potential wellbore failure in the well.

Figure 1—Three main types of cavings.

Table 1—Summary of cavings properties and failure mechanisms (Karimi 2013).


4 SPE-191634-MS

In order to automatically quantify cuttings return volume and detect cavings, Han et al., 2017, proposed
a highly portable vision-based system which employs a 2D high-resolution camera and a 3D laser sensor to
determine the physical properties of cuttings and cavings (Figure 2). The system is designed to be installed
over a cuttings ramp on a drilling rig. A combination of 2D and 3D computer vision algorithm is used to
measure cuttings and cavings physical features in real-time. The 3D point cloud/depth data obtained by this
device provides cuttings size distribution, volume and shape characteristics. The main benefits of applying
a vision-based solution are: 1) it allows comprehensive measurement of cuttings/cavings volume and
geometry; 2) the calculation is in real-time and consistent in measurement; 3) cavings can be distinguished
based on 3D point cloud /depth data analysis; 4) cavings shape analysis can help identify wellbore instability
mechanism.

Figure 2—Cuttings monitoring schematic (left), and actual cuttings slides in the field (right).

Methodology
This paper focuses on applying suitable feature extraction techniques on 3D point cloud data to achieve
cavings detection and shape categorization. In the recent decade, dense 3D point cloud data is becoming
widely applied in various industries for operational control and data analysis due to advances in 3D
sensor development. 3D object detection, however, remains challenging, especially when detecting multiple
objects of unknown shapes in a 3D scene. Due to the dense size of most 3D point cloud data, computation
time must often be sacrificed to improve algorithm accuracy. Machine learning and deep learning
approaches are the most popular techniques for 3D shape detection and recognition. Large publicly available
3D datasets (e.g. 3D ShapeNets 2018) provide a pre-labeled training set to test vision algorithm and design
a classifier. Unfortunately, drilling cuttings and cavings are not among the popular use cases for 3D vision
analysis: no 3D database of cuttings and cavings currently exist. In addition, due to the nature of drilling
operations, the sizes and shapes of cuttings and cavings often appear odd and non-uniform.
Plane segmentation is a basic 3D computer vision algorithm in the automatic reconstruction of an
unorganized point cloud data acquired by 3D scanners. One of the most common plane segmentation
approaches is the standard Random Sample Consensus (RANSAC) method by Fischler and Bolles, 1981.
This algorithm can estimate parameters of a mathematical model from a set of observed data with noise.
To achieve planar surface detection in depth data, fast normal computation using integral images can be
performed as shown in Figure 3a (Lingni Ma et al. 2013). Figure 3a shows a normal vector as a green arrow
of a plane approximated by four red point cloud coordinate points. Subsequently, normal vector calculation
is applied to the whole point cloud data (like Figure 3b). The angles between normal vectors can be grouped
based on RANSAC. By adjusting parameters like maximum point-to-point distance and maximum angular
distance, planes can be segmented from surrounding outliers. For example, Figure 3b illustrates that a table
surface (green) is distinguished from the floor (red) and table legs (white). In our algorithm, we applied a
modified plane segmentation technique from the open source 3D Point Cloud Library (PCL). After detecting
SPE-191634-MS 5

and segmenting each plane, a simple 2D convex hull polygon algorithm is applied to estimate the edges for
a set of points supported by a plane (PCL 2018). Figure 3c is an example of a 3D profile of a table surface
and a coffee mug supported by the table. The white rectangular plane (table) can be detected and marked
with red dots based on the 2D convex hull polygon algorithm.

Figure 3—Basic principle of plane segmentation and plane outline detection.

The benefit of using plane detection is the fast segmentation of larger objects in a noisy environment, such
as the occurrence of cavings among a background of cuttings in our particular situation. Cavings usually
have larger flat faces (2 to 3 times larger) than regular cuttings (Kumar et al. 2012). In order to determine the
presence cavings, the flow chart shown in Figure 4 is followed. First, plane segmentation is performed; then,
the geometrical features of detected planes are identified. Properties like Length/Thickness (L/T) ratio(φLT),
polygon geometry (angles θ, edges n), and local planes intersecting angles (Ø) are calculated. The geometric
features of each type of cavings were used as training labels for the subsequent development of a classifier
(see below).

Figure 4—Flow chart for cavings detection and geometrical features analysis.
6 SPE-191634-MS

Experimental Design
The objective of the validation experiments was to evaluate the proposed computer vision techniques on:
1) overall functionality; 2) preliminary volumetric accuracy testing; 3) feasibility of cavings detection with
shape analysis. To test the volumetric measurement accuracy and shape analysis, both uniformly shaped
objects and field cuttings were used. The purpose of testing uniformly shaped objects was to validate the
volume measurement against the detected moving speed. Experiments were conducted by sliding various
samples down a test ramp at different tilting angles (see yellow ramp in Figure 5). These experiments
confirmed the overall systems functionality and provided a baseline determination of volumetric accuracy
and 3D profile reconstruction.

Figure 5—Experimental setups.

For cavings detection and shape analysis, a speed controllable conveyor belt was employed to simulate
the sliding movement of drilling cuttings and cavings on a cuttings ramp. Since no publicly available
3D cuttings/cavings shape database exists, artificial and field collected cavings were used to collect 3D
depth point cloud data. By detecting and analyzing 3D geometry features of scanned profile, the proposed
computer vision algorithm can be evaluated for feasibility of cavings shape detection. Detected features
were trained as classifiers to allow classification of the three different cavings shapes mentioned earlier. In
order to quantify geometry features of cavings, this process requires human judgment to label dimensionless
classifiers (e.g. the length-to-thickness ratio of cavings and angularity). Once a group of classifiers was
generated and trained, a mix of cuttings and cavings was scanned to determine whether the previous defined
classifiers could detect and classify cavings. Due to the limited access to large amounts of field drilling
cavings, rocks shapes used in the experiment were selected based on their geometrical similarity to actual
cavings. The tested cavings were divided into three categories: splinter-shaped, blocky, and angular (as
shown in Figure 6). To start the experiment, each individual caving was passed through the laser profile
scanner separately. The 2D HD camera captured the moving speed of the sample and the depth profile was
measured. Then different combinations of cavings mixed with cuttings (as shown in Figures 6d - f) were
scanned and the classifier was applied to automatically detect the presence of cavings and their associated
shape. At the end, the number of cavings and their types were determined.
SPE-191634-MS 7

Figure 6—Three types of cavings samples.

Experimental Results and Discussion


Computer image processing algorithms were selected, and software was written to measure cuttings moving
speed, to analyze cuttings size distribution, to quantify volume, and to capture 3D depth of cuttings and
cavings. The laboratory testing results on volumetric calculation were previously detailed by Han et al.
(2017).
To test the capability of the proposed algorithm for cavings detection, geometric features of splinter-
shaped, tabular and angular caving samples were analyzed. Based on a top view profile scan, plane features
were first detected. Then, L/T ratio was calculated. The outlines of each plane were detected based a convex
hull polygon algorithm. The inner angles were calculated based on the detected edges. Table 1 contains
scanned and analyzed 3D profile data of caving samples.
8 SPE-191634-MS

Table 2—Plane segmentation analysis of caving samples.

Based on the orientation of scanned object, 1 to 3 faces/planes of the cavings can be detected and
distinguished (as demonstrated by sample tabular and angular cavings in Table 2). The angles between
each plane were also calculated. Figure 7 shows a scanned depth profile of a tabular caving sample semi-
immersed in the cuttings. In this particular case, it was challenging to determine the L/T ratio due to the
positioning of the tabular cavings. By analyzing the geometry of the two detected planes, inner corners were
approximated to θ = 90°±5°. The intersecting angle of the two planes is calculated as Ø= 94.23°, which is
close to a straight corner. Hence, this detected cavings was classified as tabular.
SPE-191634-MS 9

Figure 7—Tabular cavings with cuttings.

Table 3 represents scanned and calculated geometrical properties of over 100 samples representing all
three types of cavings. For a preliminary analysis, certain thresholds (φLT,θ, and Ø) were determined based
on the collected data. The L/T ratio (φLT>3.8) (generalized from Table 3 data set) was applied as the first
determining factor to distinguish splinter-shaped cavings from tabular and angular cavings. To differentiate
tabular and angular cavings, an inner corner angle of θ < 45° was used as an indication of angular shape.
The angle between adjacent planes Ø was used as an additional feature to determine the angularity of the
3D profiles. This pre-labeled training set was then applied to train a decision tree learning classifier.

Table 3—Training set of quantified geometric features of cavings.

Figure 8 is an example of how the cavings detection algorithm works. Figure 8a shows a mixture of wet
cuttings, splintery, angular, and tabular cavings. Some of the cuttings covered the tabular cavings, which
added potential difficulty to distinguish the cavings from surrounding cuttings. Figure 8b and 8c illustrates
the scanned 3D profile. After applying plane segmentation and geometrical analysis, all three cavings were
detected and their shapes were categorized as shown in Figure 8d to 8f. Figure 8e demonstrated that even
though some cuttings covered over the tabular cavings, the algorithm was still able to capture the planar
feature of the 3D data. The algorithm was able to detect cavings with minor cuttings covering on the top.
10 SPE-191634-MS

Figure 8—Cavings detection and classification.


SPE-191634-MS 11

Table 4 details the detection results when scanning mixtures of cuttings and cavings for 420 different
scenarios. Missed detection (false negative, FN) is when cavings exist but are not detected. False detection
(false positive, FP) is when the algorithm detects cavings when no cavings are present. Correct detection
(true positive, TP) is when the algorithm correctly detects the presence of cavings. No detection (true
negative, TN) is when the system detects no cavings when none exist in the scene. N is the number of
cavings correctly classified as splinter-shaped, tabular and angular. The cavings detection accuracy and
classification accuracy is calculated as:

The low false detection rates in Table 4 demonstrate that the algorithm is sufficiently accurate in
distinguish cavings from cuttings. The false detection was generally caused by cuttings piling up together.
Missed cavings (FN) were largely due to cavings being fully covered under cuttings. Overall, the algorithm
detected the presence of cavings in 88% of all the test cases. With regards to cavings categorization, splinter-
shaped cavings are more easily classified because of their unique geometrical features (particularly L/T
ratio). In our study, tabular and angular cavings had a lower classification accuracy (approximately 70%).
One of factors causing this lower accuracy is that tabular and angular cavings are geometrically similar and
it is different for even humans to visually classify them correctly. Also, when tabular and angular cavings
are partially covered by cuttings, their geometrical properties are much more difficult to quantify.

Table 4—Cavings detection and classification.

The accuracy of cavings shape categorization is to a large extent based on the pre-scanned training set.
The accuracy of the trained classifier can be improved by scanning more samples of field-collected cavings.
Other machine learning approaches such as support vector machines can also be applied to improve the
accuracy of shape categorization.
A preliminary field trial was performed at a shale shaker testing facility to evaluate the system accuracy
and robustness in an outdoor environment. Since the facility was not on an actual rig, the sensing system
could not be mounted below the floor level of the shale shaker (as suggested in Figure 2). A makeshift ramp
was installed right below the outlet of the shale shaker screen, and the 2D camera and 3D laser scanner
were set above this ramp where cuttings slide down after being separated from the drilling fluid. The blue
machine in Figure 9 is a typical shale shaker system. Sand was used to simulate drilling cuttings and water-
based drilling fluid was circulated. Before turning on the shale shaker system, the laser scanner and the HD
camera were calibrated on a clear ramp. The movement of the cuttings was kept constant by using a constant
drilling fluid flow rate. A half pipe was used to collect wet cuttings and the volumetric flow rate was thus
calculated. The scanner and camera were scanning and collecting the data at a pre-set high frequency. As
cuttings slid down the ramp at a relatively steady pace, an average cuttings moving speed was calculated
12 SPE-191634-MS

every 15s (as shown in Figure 10). By merging the detected speed and 3D depth data, the volume of cuttings
was calculated and the 3D reconstructed profiles were stored for cavings detection and shape analysis. The
purpose of this preliminary backyard testing was to evaluate the sensing system in an outdoor environment
and prepare the sensors for actual field testing.

Figure 9—Field test setup at NOV shale shaker facility.


SPE-191634-MS 13

Figure 10—Flow chart for cuttings volumetric measurement and 3D profile scanning.

This preliminary testing successfully demonstrated that the proposed design is able to capture the physical
properties of cuttings/cavings in an outdoor setting. Figure 11a shows the scanned data that was analyzed to
calculate the volume and 3D geometry of cavings in real-time. Figure 11b shows image capturing of a tabular
caving sample among the sand cuttings. The system furthermore calculated cuttings volumetric flow with an
error of + 7.74%. Some factors that may have affected the accuracy are the variances in cuttings movement
speed measurement and the over-calculated volume due to void spaces under the cuttings / cavings. More
field testing will be performed to address these potential problems. In addition, with the benefit of real-time
3D depth data, a reconstructed cuttings profile can also be used to monitor shale shaker performance. For
instance, Figure 11c shows a scenario where the cuttings from the shale shaker was relative dry and the
presence of drilling fluid was low. However, when the viscosity of the drilling mud was higher, the shale
shaker failed to effectively separate the cuttings and the drilling fluid efficiently. This phenomenon was
captured in Figure 11d, where excess drilling fluid was observed. Algorithms can be designed to capture
such events automatically and alarm the mud logger to check shale shaker performance.
14 SPE-191634-MS

Figure 11—Field testing results at shale shaker yard testing facility.

Conclusions and Future Work


In this paper, a novel 3D depth-sensing system is introduced and evaluated to measure cuttings volume,
identify cavings, and to analyze the shape of the cavings. The proposed sensing system consists of a
unique combination of 3D and 2D sensing technologies, which allows for non-intrusive real-time analysis
of cuttings and cavings to monitor hole cleaning efficiency and wellbore instability. A robust plane
segmentation computer vision algorithm was applied to first extract geometrical properties of the cavings.
The extracted geometrical features were then used with a training set to design a classifier to categorize
cavings shapes. The proposed 3D shape analysis algorithm achieved a cavings detection accuracy at
about 90% and a 75% accuracy in recognizing the shape of detected cavings. Preliminary field testing
demonstrated the capability of sensing algorithm to measure cuttings return volume and detect cavings in
real-time. Based on the real-time reconstructed 3D profiles, the system can also examine the shale shaker
performance by detecting sudden change of the cuttings profile pattern.
Overall, this state-of-art, non-intrusive computer visualization system evaluates cuttings/cavings
quantitatively and helps monitor downhole conditions to reduce drilling-related NPT in the field. By
diagnosing potential hazards based on the shapes of cavings, the system warns the driller on adverse wellbore
failure and instability conditions and the likelihood of associated operational problems and non-productive
time.
SPE-191634-MS 15

Acknowledgement
The authors would like to thank ConocoPhillips for financial support for this project and providing field
assistance during the test stages. We would also like to acknowledge National Oilwell Varco (NOV)
for assisting in the project and allowing testing at their facility. The sponsors and members of the Rig
Automation and Performance Improvement in Drilling (RAPID) at the University of Texas at Austin are
also thanked for their support and assistance through this study.

References
A.M., Ramadan. 2001. Mathematical Modeling and Experimental Investigation on Solids and Cuttings Transport. Ph.D.
Dissertation, Norwegian University of Science and Technology.
Aldred, W., D. Plumb, I. Bradford, L. Cousins, J. Fuller, S. Goraya, and Tuker D. 1993. Oilfield Review - Managing
Drilling Risk. Review, Schlumberger.
Azar, Jamal J., and Robello Samuel. 2007. Drilling Engineering. PennWell Books.
Baldino, S., R. E. Osgouei, E. Ozbayoglu, S. Miska, N. Takach, R. May, and D. Clapper. 2015. "Cuttings Settling and
Slip Velocity Evaluation in Synthetic Drilling Fluids." Offshore Mediterranean Conference and Exhibition. Ravenna,
Italy: Offshore Mediterranean Conference.
C. Aldea, A. W. Iyoho, M. Zamora. 2005. "Hole Cleaning: the Achilles’ Heel of Drilling Performance?." Houston, Texas:
American Association of Drilling Engineers.
C.J. Hopkins, R.A. Leicksenring. 1995. "Reducing the Risk of Stuck Pipe in The Netherlands." Amsterdam: Society of
Petroleum Engineers.
Cayeux, Eric, Taiwo Mesagan, Sakti Tanripada, Mohamed Zidan, and Kjell Kåre Fjelde. 2014. "Real-Time Evaluation
of Hole-Cleaning Conditions With a Transient Cuttings-Transport Model." SPE Drilling & Completion. Society of
Petroleum Engineers. doi:https://doi.org/10.2118/163492-PA.
Chien, Sze-Foo. 1972. "Annular velocity for rotary drilling operations." International Journal of Rock Mechanics and
Mining Sciences & Geomechanics Abstracts 9 (3): 403–416. doi:https://doi.org/10.1016/0148-9062(72)90005-8.
Clark, R.K., and K.L. Bickham. 1994. "A Mechanistic Model for Cuttings Transport." SPE Annual Technical Conference
and Exhibition. New Orleans: SPE. doi:http://dx.doi.org/10.2118/28306-MS.
Cranford, P.J., Muschenheim, D.K., Loder, J.W., Gordon, D.C. Jr, Keizer, P.D., and Kranck, K.,. 1992. Predicting the
environmental impacts of drilling wastes on Georges Bank scallop populations. ICES Benthos Ecology Working
Group.
Ferranod, Pierrick, and Slim Hbaieb. 2015. "Schlumberger CLEAR Hole Cleaning and Wellbore Risk Reduction Service."
Accessed 09 03, 2016. http://www.slb.com/~/media/Files/drilling/industry_articles/2015-clear-ep.pdf.
Gang Han, Jo Henson, Andy Timms, Ilyah Abdul Aziz. 2009. "Wellbore Stability Study: Lessons and Learnings from
Technically Active Field." Asia Pacific Oil and Gas Conference & Exhibition. Jakarta, Indonesia: Society of Petroleum
Engineers.
Gavignet, A.A. and Sobey. 1989. "Model Aids Cuttings Transport Predictions." Journal of Petroleum Technology (Society
of Petroleum Engineers) 41 (09). doi:http://dx.doi.org/10.2118/15417-PA.
Han, R., Ashok, P., Pryor, M., Oort, E. van, Scott, P., Reese, I., & Hampton, K. 2017. "Real-Time Borehole Condition
Monitoring using Novel 3D Cuttings Sensing Technology." SPE/IADC Drilling Conference and Exhibition. The
Hague, The Netherlands: Society of Petroleum Engineers. doi:10.2118/184718-MS.
IPIECA. 2009. "Drilling fluids and health risk management." International Petroleum Industry Environmental
Conservation Association. http://www.ogp.org.uk/pubs/396.pdf.
Jalukar, Laxmikant Shrihari. 1993. A study of hole size effect on critical and subcritical drilling fluid velocities in cuttings
transport for inclined wellbores. Thesis/dissertation, Unversity of Tulsa.
Karimi, Moji. 2013. "Drill-Cuttings Analysis for Real-Time Problem Diagnosis and Drilling Performance Optimization."
SPE Asia Pacific Oil and Gas Conference and Exhibition, 22-24 October. Jakarta, Indonesia: Society of Petroleum
Engineers. doi:http://dx.doi.org/10.2118/165919-MS.
Kumar, S. Ansari, S. Wang, J. Yiming, S. Ahmed, M. Povstyanova, and B. Tichelaar. 2012. "Real-time Wellbore Stability
Analysis An Observation from Cavings at Shale Shakers." AAPG International Convention and Exhibition. Singapore:
AAPG.
Larsen, T.I., A.A. Pilehvari, and J.J. Azar. 1997. "Development of a New Cuttings-Transport Model for High-Angle
Wellbores Including Horizontal Wells." SPE Drilling & Completion (Society of Petroleum Engineers) 12 (02).
doi:https://doi.org/10.2118/25872-PA.
16 SPE-191634-MS

Li Y., Bjorndalen N., Kuru E. 2007. Numerical Modeling of Cuttings Transport in Horizontal Wells Using Conventional
Drilling Fluids. Canadian International Petroleum Conference, Calgary, Canada: Alberta and Journal of Canadian
Petroleum Technology.
Lingni Ma, Raphael Favier, Luat Do, Egor Bondarev, Peter H. N. de With. 2013. "Plane segmentation and decimation of
point clouds for 3D environment reconstruction." IEEE 10th Consumer Communications and Networking Conference
(CCNC). Las Vegas, NV, USA: IEEE. doi:10.1109/CCNC.2013.6488423.
M. A. Fischler, R. C. Bolles. 1981. "Random Sample Consensus: A Paradigm for Model Fitting with Applications to
Image Analysis and Automated Cartography." Comm. of the ACM 24: 381–395.
Miska, Stefan, Troy Reed, Ergun Kuru, Nicholas Takach, Kaveh Ashenayi, Ramadan Ahmed, Mengjiao Yu, and et al.
2004. Advanced Cuttings Transport Study. Final Technical DOE Report, Tulsa, Oklahoma: The University of Tulsa.
doi:DE-FG26-99BC15178.
Moore, L.P. 1974. In Drilling Practices Manual, Chap. 8. Tulsa: Petroleum Publishing Co.
Naegel, M., E. Pradie, T. Delahaye, C. Mabile, and G. Roussiaux. 1998. "Cuttings Flow Meters Monitor Hole Cleaning
in Extended Reach Wells." European Petroleum Conference, 20-22 October. The Hague, Netherlands: Society of
Petroleum Engineers. doi:http://dx.doi.org/10.2118/50677-MS.
Nguyen, D., and S. S. and Rahman. 1996. "A Three-Layer Hydraulic Program for Effective Cuttings Transport and Hole
Cleaning in Highly Deviated and Horizontal Wells." IADC Asia Pacific Drilling Technology. Kuala Lumpur, Malaysia:
SPE.
Osisanya, Samuel O. 2012. Practical Approach to Solving Wellbore Instability Problems. Society of Petroleum Engineers.
PCL. 2018. "Construct a convex hull polygon for a plane model." Point Cloud Library. Accessed 04 17, 2018. http://
pointclouds.org/documentation/tutorials/convex_hull_2d.php.
Pigott, R.J.S. 1941. "Mud Flow In Drilling." Drilling and Production Practice 91.
Princeton Vision & Robotics. 2018. 3D ShapeNets: A Deep Representation for Volumetric Shapes. Accessed 05 14, 2018.
http://3dshapenets.cs.princeton.edu/.
Rubiandini, Rudi. 1999. "Equation for Estimating Mud Minimum Rate for Cuttings Transport in an Inclined-Until-
Horizontal Well." Abu Dhabi: Society of Petroleum Engineers.
Salazar-Mendoza, R., and A. Garcia-Gutierrez. 2008. "A Two-Region Hydraulic Averaging Model for Cuttings Transport
During Horizontal Well Drilling." Journal of Canadian Petroleum Technology (Petroleum Society of Canada).
Santana, A.L. Martins | C. Costapinto. 1992. "Evaluation of Cuttings Transport in Horizontal and Near Horizontal Wells-
A Dimensionless Approach." SPE Latin America Petroleum Engineering Conference. Caracas, Venezuela: Society of
Petroleum Engineers. doi:https://doi-org.ezproxy.lib.utexas.edu/10.2118/23643-MS.
Wilson, K.C. 1976. "A Unified Physically-Based Analysis of Solid-Liquid Pipeline Flow." 4th International Conference
on the Hydraulic Transport of Solids in Pipes. Golden.
Yongfeng Kang, Mengjiao Yu, Stefan Z. Miska, Nicholas Takach. 2009. "Wellbore Stability: A Critical Review and
Introduction to DEM." SPE Annual Technical Conference and Exhibition. New Orleans: Society of Petroleum
Engineers. doi:https://doi-org.ezproxy.lib.utexas.edu/10.2118/124669-MS.
Zeidler, Udo H. 1972. An Experimental Analysis of the Transport of Drilled Particles. Thesis, Houston: Rice University.

Вам также может понравиться