Вы находитесь на странице: 1из 19

SPE/IADC Drilling Conference and Exhibition 2017

All manuscripts will be sent through an XML tagging process that will standardize
the look of the paper and create links for figures, equations, and references.
Figures and tables should be placed directly after the first paragraph they are
mentioned in. The XML tagging will not alter the technical content of the paper.

SPE-184615-MS

A Real-Time Well-Site Based Surveillance and Optimization Platform for


Drilling: Technology, Basic Workflows and Field Results
G. S. Payette, B. J. Spivey, L. Wang, ExxonMobil Upstream Research Company; J. R. Bailey, ExxonMobil
Development Company; D. Sanderson, XTO Energy; R. Kong, M. Pawson, A. Eddy, Pason Systems

Copyright 2017, Society of Petroleum Engineers

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 reprod uction, 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
We describe recent technology advances, basic workflows and field trial results for a real-time well-site
based surveillance and optimization software platform for drilling. The system integrates with existing
infrastructures available at a typical rig-site and may be deployed in the driller’s cabin to enable real-
time decisions by drilling personnel. The platform leverages as input one-second real-time surface data
(obtained from sensors instrumented on surface equipment and available at most well sites). The
platform provides surveillance capabilities, trending analysis tools, and controllable drilling parameter
recommendations to improve drilling performance by enhancing decision-making at the well-site.
The platform supports active drilling parameter management by encouraging regular drill-off tests.
Understandings obtained from these tests allow the system to provide real-time information on
performance trends and drilling dysfunctions through various displays which aid in the drilling
optimization process. We describe recent technology enhancements to the system which leverage
adaptive response surface technologies that map out performance and dysfunction for the controllable
drilling parameters as drill-off tests are performed. The system allows users to view performance maps
for a variety of system outputs including Mechanical Specific Energy (MSE), Rate of Penetration
(ROP), stick-slip severity, and Depth of Cut (DOC), as well as combinations of these outputs. The
performance maps give precedence to “new” data and adapt as the bit wears and/or new formations are
encountered and are leveraged to provide controllable drilling parameter recommendations to the driller.
We present examples from recent field applications of the system which demonstrate the value of
real-time well-site based drilling optimization and parameter management leveraging surface data. We
provide examples where the software was able to provide operation personnel with the confidence to
increase Weight on Bit beyond “field rules” to enable faster drilling operations as compared with offsets.
We provide additional examples highlighting how active surveillance and parameter management using
the system’s recommendations and response surface maps was able to produce multiple record bit runs
in mature fields. We also discuss workflows enabled by the system and efforts taken to enhance field
personnel uptake by working to address the human machine interaction aspect of the platform during
software development.
2 SPE-184615-MS

Introduction
Operators and service companies continue to innovate and develop technologies aimed at reducing the
cost of well construction. One area with significant opportunities for improving drilling performance,
accelerating the “learning curve” and enabling consistency in drilling operations lies in better utilization
of the real-time data that is collected at well sites. Just a few examples from the 2016 SPE/IADC
Drilling Conference and Exhibition detailing efforts taken to leverage real-time drilling data to improve
operational performance include: (Greenwood 2016), (Teelken et al. 2016), (Jiang and Samuel 2016),
(Mirani and Samuel 2016), (Evangelatos and Payne 2016), (Nabavi et al. 2016), (Schils et al. 2016),
(Salminen et al. 2016) and (Kristjansson et al. 2016). Although there are many opportunities for
leveraging real-time data, one example with significant potential is the implementation of a software
platform designed to leverage “surface drilling data” measurements to enable better decision making at
the well-site as part of the Rate of Penetration (ROP) optimization / management process.
Surface drilling data, i.e., data measurements obtained from sensors instrumented on equipment
located at the surface, is extremely useful as it is almost always available on location (both onshore and
offshore), may be transmitted using industry standard communication protocols (such as the Wellsite
Information Transfer Specification or WITS) and when appropriately processed can be used to infer a
great deal about drilling performance. Two quantities which can be computed directly using surface
data that can provide insight into drilling performance include Mechanical Specific Energy (MSE) and
the operator’s Torsional Severity Estimate (TSE). A brief description of each of these quantities is
provided below:
 Mechanical Specific Energy (MSE): A measure of energy expended by a drilling assembly
while drilling an interval of rock (normalized by the volume of rock drilled). MSE correlates
with compressive rock strength under efficient drilling conditions and tends to increase in the
presence of dysfunction (such as bottom hole assembly or BHA lateral vibrations, bit balling,
drill string buckling, etc.). As a result, MSE may be used as a dysfunction indicator while
drilling. A variety of formulas for evaluating MSE can be found in the literature including the
original equation presented by Teale (Teale 1965) and generalizations for use with mud motors,
hydraulics, etc.
 Torsional Severity Estimate (TSE): An estimate of stick-slip severity or the magnitude of
torsional vibrations at the bit. This quantity may be determined based on the surface rotary
speed (RPM), characteristics of the surface torque signal and a simple frequency domain based
mechanics model of the drilling assembly [see for example (Ertas et al. 2014) and (Chang et al.
2014)]. The wave propagation properties of drill pipe are such that there is typically very little
time delay (typically a few seconds or less) between the occurrence of torsional events at the bit
and their manifestation in measured torque at the surface.
MSE can be viewed as a quantity with relative, as opposed to absolute, meaning as the baseline for
minimum expected MSE (no dysfunction) varies with changes in lithology. TSE on the other hand may
be viewed as an absolute measure in the sense that a value of 1 always denotes a state of full stick-slip.
An effective ROP management process for drilling a given formation section should ideally include
frequent investigations of the controllable drilling parameter space via controllable drilling parameter set
point changes, where the controllable drilling parameters typically include Weight on Bit (WOB) and
RPM. Additional controllable parameters sometimes include ROP and for mud motor drilling
operations Differential Pressure (Diff P). These investigations (sometimes referred to as a sequence of
“drill-off tests”) may then be used to determine optimal set points for the controllable drilling
parameters which ideally lead to high ROP, relatively low MSE (low lateral vibrations, bit balling, etc.)
and low TSE (low stick-slip). It is important to note that at times other constraints (ECD, hole cleaning,
surveys, etc.) will take precedence. The ultimate goal of an effective ROP management progress is
consistently better and longer bit runs.
SPE-184615-MS 3

Unfortunately, the investigations needed to assess drilling performance can be difficult to implement
in the field by drilling personnel. They require a statistically meaningful sample of multiple
combinations of set points for the controllable drilling parameters and proper identification of trends in
the calculated drilling performance data (ROP, MSE, TSE, etc.) while also giving credence to changing
drilling conditions.
To address many of these challenges, a real-time well-site centric drilling surveillance and
optimization software platform has been developed. The system leverages surface drilling data
measurements and may be deployed in the driller’s cabin to enhance the ROP optimization process. The
platform provides surveillance capabilities, trending analysis tools, and controllable drilling parameter
recommendations to help optimize drilling performance. The system is intended to remove the more
burdensome aspects of the ROP management process from the driller’s workflow.

Overview of Technology
In this section we provide a brief technical overview of the surveillance and optimization platform for
drilling. The system has been in development for several years and some technical aspects have been
reported previously in the literature; for example, see (Chang et al. 2014) and (Payette et al. 2015). In
this section we summarize key technical aspects reported by (Payette et al. 2015) and highlight
additional system capabilities that have been developed in the interim.
The platform is a software application deployed in the driller’s cabin that possesses automated real-
time data acquisition, vibration analysis capabilities, advanced trending analysis tools, guided drill-off
test workflows and various displays that enable optimization of the drilling process. The platform is
currently an advisory system only, i.e., no closed-loop control. The focus of the system is on displaying
pertinent information relating to drilling performance and controllable drilling parameter management
that best enables decisions and actions to be taken by operations personnel at the well-site. An overview
of the stages of interaction between the system, other systems on location (i.e., the mud logging system
to obtain WITS data) and the driller are shown in Table 1.

Table 1 – Stages of interaction between the surveillance and optimization platform (for data acquisition, analysis and display) and
the driller (for action implementation). Table format adapted from (Parasuraman, Sheridan, and Wickens 2000) and (de Wardt,
Sheridan, and DiFiore 2016).

Data Acquisition Analysis Display Actions


Surface data – WITS Rec. 1 State estimation: Buttons showing current / System setup / configuration
WOB, surface RPM, surface On-bottom or off-bottom, instantaneous values
TQ, block height, etc. rotary drilling or sliding Auto Driller control mode
Strip charts showing selection
Short time / depth scale: historical values
Filtered drilling perform- Auto Driller set point
ance data – ROP, MSE, Stick-slip alarm changes (WOB, Diff P,
DOC, TSE, etc. ROP, RPM)
A calibration progress bar
Medium time / depth scale: System calibration /
Response points, calibration Heat maps / surfaces recalibration
points showing drilling
performance as a function of Modify (expand / reduce)
Long time / depth scale: the controllable drilling bounds for acceptable
Multi-variable regression / parameters controllable drilling
surface fitting, parameters ranges
recommendations Recommendations
4 SPE-184615-MS

Data Acquisition
The system uses data obtained from sensors instrumented on surface equipment as inputs to various
algorithms for surveillance and optimization of the drilling process. Data acquisition by the system is
obtained from the mud logger’s electronic data recording equipment using the industry standard Wellsite
Information Transfer Specification (WITS). The surveillance and optimization platform currently relies
exclusively on data transmitted in WITS Record 1 which includes surface data measurements such as
hookload, weight on bit, rotary speed, standpipe pressure, block position, bit depth, hole depth, etc.
Additional non-WITS Record 1 standard surface measurements, such as differential pressure, may be
configured through the use of spare channels in Record 1. Data acquisition is required at least once a
second to ensure the ability of the system to perform stick-slip surveillance analysis. Simple outputs
from the system may be communicated back to the mud logging system through a custom WITS record
and streamed to an off-site server for remote viewing by engineering staff through a web-based
application.

Data Analysis
The system offers a variety of analysis capabilities, a high level summary of which is provided in Table
1. The most elementary level of analysis is basic cleanup of the raw data acquired through WITS
Record 1. Linear interpolation is applied to remove null values and median-order filtering of a select
number of data channels is invoked to eliminate data spikes; see (Payette et al. 2015) for additional
details. Simple state estimation is performed to determine whether the bit is on-bottom or off-bottom
and whether the assembly is rotary drilling or sliding (mud motor case with no RPM applied at surface).
The additional (primary) levels of processing and computation performed by the system can be broadly
categorized into: (a) short time / depth scale analyses, (b) medium time / depth scale analyses and (c)
long time / depth scale analyses.

Short time / depth scale analysis


The short time / depth scale analysis is associated with determining basic measures of drilling
performance. These measures require appropriate filtering / averaging with respect to windows in depth
and / or time. There are two types of calculations performed at this level: those whose averaging
windows are predominately dictated by changes in Measured Bit Depth (MD) and stick-slip estimation
using surface data which requires averaging / windowing with respect to time.
Among the drilling performance measures that are largely associated with changes in measured bit
depth (MD) are: Rate of Penetration (ROP), Mechanical Specific Energy (MSE), Depth of Cut (DOC)
and Bit Aggressiveness / Bit Friction Coefficient (μ) which may be defined as
d Bit depth
ROP  (1)
dt
1  TQ[klbs  ft] 
MSE[ksi]  2 
WOB[kip]  24  (2)
A[in ]  DOC[in/Rev.] 
ROP[ft/hr]
DOC[in/Rev.]  (3)
5RPM[Rev./min]
36  TQ[klbs ft]
[in/Rev.]  . (4)
WOB[klbs]  Dia[in]
The surveillance and optimization platform adopts modifications of the above formulas for MSE and
DOC for mud motor drilling environments which require as input basic mud motor properties (to
estimate mean down hole torque and rotary speed).
In practice, the “continuous” formulas above must be approximated through averaging procedures
which are robust in very slow and very fast drilling environments. Our approach is to adopt depth-based
SPE-184615-MS 5

averaging to manage the trade-offs between the competing objectives of: (a) minimizing Noise to Signal
(largely driven by errors associated with estimating changes in MD) and (b) avoiding excessive over-
smoothing. In practice ROP, MSE and DOC are evaluated as
Bit depth  ti ; ti  P 
ROP  ti ; ti  P   (5)
t  ti ; ti  P 
1  TQ 
MSE  ti ; ti  P   
ti
 WOB  24  ROPdt (6)
ti P Asi  DOC 
ROP  ti ; ti  P 
DOC  ti ; ti  P   (7)
5 Average RPM  ti ; ti  P 
where i is a sampling index for time. The above formulas constitute estimates for ROP, MSE and DOC
at time ti that have been generated using averaging procedures with respect to the time window [ti – P, ti]
where P > 0 and ti > ti – P. The change in bit depth may be estimated using the change in block height
(∆BH) and a simple mechanics model of the drill string that compensates for stiffness effects (pipe
stretch) and filtered changes in hookload (∆HL)
HL  ti ; ti  P 
Bit depth  ti ; ti  P   BH  ti ; ti  P   . (8)
  MDi 
In the above formula  is the effective spring constant of the drilling assembly at the current measured
depth. The time window [ti – P, ti] is constructed to ensure that Bit depth  ti ; ti  P   s , where s is a
user prescribed (constant) depth increment on the order of 0.5 to 1 ft. Figure 1 highlights the need to
account for drill string compliance when computing depth-based quantities such as ROP. This is
especially important to compensate for effects associated with changes made to the Auto Driller set
point for weight on bit.

Figure 1 – Strip charts showing the relationship between the controllable drilling parameters (WOB and RPM) and ROP. These
traces highlight the importance in correcting for pipe stretch/compression when computing ROP based on the block height.
6 SPE-184615-MS

Stick-slip severity is estimated by the system using measured surface torque, RPM and frequency-
domain based dynamics model of the drilling assembly (Ertas et al. 2014). The model requires basic
rheology properties of the drilling fluid and a simple drill string description to determine theoretical
values for the stick-slip period T and the surface torque swing at full stick-slip per RPM or ∆TQref. Both
quantities are functions of MD and may be used to estimate stick-slip severity in real-time using the
following formula
  
max TQi , TQi 1 ,..., TQi Q  min TQi , TQi 1 ,..., TQi Q 
 
TSE ti ; ti Q  (9)

TQ ref  MDi   median RPM i , RPM i 1 ,..., RPM i Q 
where the time window [ti – Q, ti] is taken such that ti – ti – Q > T. TSE is defined such that a value of 0 is
associated with no torsional vibrations and a value of 1 is indicative of a state of full stick-slip. This
stick-slip estimate may be used to compute a downhole RPM envelope (min / max RPM) as shown in
the rotary (non-motor) drilling example in Figure 2. The system can also compute a bit TSE value for
motor drilling environments which requires the above formula to be multiplied by the ratio of the
surface and mean bit rotary speeds. It is important to note that in general the time windows for stick-slip
monitoring are much shorter than the time / depth windows used to compute depth-based quantities such
as ROP and MSE. In general, there is little delay between the occurrence of torsional events down hole
(e.g., small / medium levels of rotational vibrations, full / severe stick-slip or bit sticking events) and
their manifestation in the system’s computed stick-slip signals.

Figure 2 – Stick-slip monitoring using surface TQ and RPM and a mechanics model of the drill string.

Medium time / depth scale analysis


The medium time / depth scale analysis is associated with processing the filtered data obtained
through the short time / depth scale analysis to produce quantities that can be used for optimization (long
time / depth scale analysis). We refer to these basic quantities as “response points” and “calibration
points”. A “response point” is a depth-based quantity associated with a small (typically ½ to 1 ft)
interval of drilled rock that includes “average” values for each of the controllable drilling parameters
and the drilling performance data obtained by the short time / depth scale analysis. Response points are
non-overlapping in time and depth and are generated by the system whenever the controllable drilling
parameters are detected to be steady. A response point can be viewed as a set of average controllable
drilling parameters (such as WOB and RPM) and filtered drilling variables (such as ROP, MSE, DOC
and TSE).
A “calibration point” is a collection of neighboring response points determined to be sufficiently
close in the 2-D controllable drilling parameter space. The system considers the following 2-D control
modes: WOB / RPM, ROP / RPM and Diff P / RPM. By definition, a calibration point is required to
SPE-184615-MS 7

contain a minimum, typically no less than 4, number of response points. Calibration points are used by
the system to identify regions of the controllable drilling parameter space which have been sampled
sufficiently to establish meaningful mean values for the drilling performance measures. Calibration
points are identified using Euclidean-based clustering algorithms that leverage user defined “step sizes”
for the controllable drilling parameters. For weight on bit control, step sizes of 2 klbs and 5 RPM are
commonly adopted by the system.

Long time / depth scale analysis


The long time / depth scale analysis is used to construct statistical models (response surfaces) relating
the controllable drilling parameters to the drilling performance variables. Processing of these models is
further utilized to produce recommended set points for the controllable drilling parameters.
The response surfaces are constructed using advanced regression techniques, which generate smooth
surfaces for each of the drilling performance variables as a function of the controllable drilling
parameters. Response surface fitting is performed using all response points associated with the
identified calibration points. For example, for N response points meeting the above criteria, a response
surface may be constructed for a given drilling performance variable u(x), with given values ui sampled
at controllable drilling parameter locations xi (where i ranges from 1 to N) by minimizing an
appropriately discretized version of the following functional with respect to u
1 N 
  u  x      i  u  x   ui    x  xi   d    2u  x  .
2 2
(10)
2  i 1   2  ,0

In the above expression Ω is the convex controllable drilling parameter domain, λi are a set of weights,
δ(x – xi) is the so-called delta-function, γ is a regularization parameter and || ||Ω,0 denotes the standard
L2(Ω) norm. The weights λi are automatically prescribed by the system using an exponential decay
function which down-weights or ages-out “older” response points which may not be as relevant due to
potential changes in lithology. Examples of response surfaces for MSE, ROP and TSE are shown in
Figure 3. In this figure, the response points are the small black dots and the calibration points are the
green cubes; note that the surfaces extrapolate out a small amount beyond the sampled response point
data.
In order for the system to make recommendations, an objective function (OBJ) or objective surface is
necessary. The purpose of the objective surface is to combine multiple (potentially competing) drilling
performance objectives into a single objective for drilling performance. In the most general case, the
surveillance and optimization system generates OBJ as a combination of “drilling efficiency”, ROP and
TSE. The system allows the user to select either: surface MSE, downhole MSE (mud motor) or DOC /
WOB as the “drilling efficiency” parameter. In general the drilling performance measures used for
optimization depend on the Auto Driller control mode and the operation mode as shown in Table 2. An
example OBJ that has been used to handle the most general case is given by the following formula
max 0, 1   TSE n  
ROP
OBJ  (11)
MSE
where α and n are positive real valued numbers whose values are set to severely penalize the OBJ for
cases with high stick-slip. We note that other OBJ formulas are also possible. An example OBJ surface
is shown in Figure 3. In this example the high stick-slip region is highly penalized in OBJ.
Generating recommendations is the final step in the “long time / depth scale analysis”.
Recommendations are largely based on trends identified in the temporally evolving OBJ response
surface. However, when the surveillance and optimization platform is initialized or recalibrated,
recommendations are initially somewhat ad hoc and are provided to obtain sufficient data for drilling
performance trending analysis.
8 SPE-184615-MS

Figure 3 – Individual response surfaces for MSE, ROP and TSE. These surfaces form the basis for the objective function (OBJ).

Table 2 – Drilling performance variables utilized to optimize drilling performance for an operating mode and a user selected Auto
Driller control mode. The “efficiency” parameter is either: surface MSE, down hole MSE (mud motor) or DOC / WOB.

Drilling Performance Variables


Operation mode Control mode Efficiency ROP TSE
Rotary drilling WOB   
Rotary drilling Diff P   
Rotary drilling ROP  – 
Sliding (motor) WOB   –
Sliding (motor) Diff P   –
Sliding (motor) ROP  – –

Human-Machine Interaction (HMI), Workflows


The optimization and surveillance platform ultimately requires decisions and actions to be taken by the
driller. The need to address Human-Machine Interaction (HMI) was, therefore, identified early in the
development of the software. The goal of the HMI component was to create a simple user interface for
the system that the driller can understand with minimal training.
A design studio exercise was conducted at a well site to engage drilling operations personnel (i.e.,
drillers, tool pushers, rig supervisors, etc.) in a brainstorming exercise to arrive at ideas for a simple
User Interface (UI) design. The basic ideas for such an exercise can be summarized as follows (an
illustration is shown in Figure 4):
SPE-184615-MS 9

 In 10 minutes of taking pencil to paper, a group of 10-15 people come up with 100+ ideas.
 Over the course of several brainstorming sessions ideas are generated, filtered, re-mixed,
evaluated, and iterated upon until a single testable concept emerges.
The final concept may be subsequently evaluated for production cost, tested by a usability team, and
presented to customers for approval / feedback.

Figure 4 – A design studio is a brainstorming exercise where a conceptual solution to a problem is created using pencil sketches
and crowd sourcing.

Two prototype UI views (developed based on the design studio exercise and field testing) are shown
in Figure 5. Once configured and initiated, the system operates by continually performing the basic
analysis steps summarized in Table 1 and providing information to the driller. The driller interacts with
the system in few different ways including:
 Performing configuration / system setup which requires entering a simple drill string description,
mud motor parameters (if applicable), hole size, min / max ranges for the controllable drilling
parameters and selecting an Auto Driller control mode (WOB, ROP or Diff P).
 Implementing Auto Driller set point changes based on information gained from visual trends
identified from the surfaces / heat maps or recommendations obtained from the system.
 Calibrating the system as necessary. Calibration is performed by pressing a button which clears
system memory. This action is typically performed upon encountering new formations.
The system initially begins on startup or following initiation of calibration in a learning or calibration
mode (see left image in Figure 5). This mode is associated with a system state with little data on which
to perform optimization. During calibration mode, the goal of the system is to obtain enough data to
sufficiently sample the controllable drilling parameter space to assess the performance landscape.
Currently, the system requires the user to obtain 5 calibration points to complete system calibration (this
number is somewhat ad hoc and could be adjusted in the future). During the calibration stage a progress
bar is shown on the top left hand corner of the UI to indicate the number of calibration points which
have been obtained. The progress bar also shows a percentage complete value for the most recent (but
not yet completed) calibration point. Once calibration points are obtained, the surface fitting routines
are run to generate heat maps of drilling performance. The user may view maps for drilling performance
variables such as MSE, ROP, stick-slip, DOC, or the OBJ (labeled as “Summary”). The heat maps are
colored green (better) to red (worse) on a relative scale for all drilling performance outputs with the
exception of TSE which is colored on an absolute basis.
10 SPE-184615-MS

Figure 5 – Prototype User Interface screenshots of the system showing the system in: (a) “calibration mode” and (b) “optimization
mode”. In each image, the “bit” is positioned at the current WOB and RPM and the “star” is located at the recommended WOB and
RPM.

Throughout the calibration mode, the system provides recommendations to the user. The driller may
proceed through the calibration phase through any combination of either following the system
recommendations or simply making periodic set point changes that lead to the generation of calibration
points throughout the controllable drilling parameter space. Initially recommendations are made in an
ad hoc fashion and in later stages (after obtaining three calibration points) using trends obtained from an
analysis of the OBJ surface.
Upon completion of the calibration mode, the system transitions into an optimization mode (see right
image in Figure 5) where the performance maps are updated continually but new recommendations are
only made at a frequency which is prescribed by the driller. Notice that the progress bar is absent during
the “optimization phase”. An example of running the system against simulated data is shown in Figure
6. In this example the user proceeds through calibration and eventually enters optimization mode.
During the optimization phase, the driller lowers the RPM and encounters a region of severe stick-slip
(TSE > 1). The system then helps to work the driller out of stick-slip by recommending raising the
rotary speed (see bottom three images of Figure 6).
Communication between the well-site and the office (engineering / operations support) is necessary
to fully maximize value when running the system in the field. Among the items which often need to be
relayed from the office to the field include:
 Which drilling operations (rotating vs. sliding for directional work), hole sections (surface,
intermediate, production) and formations are deemed most appropriate to consider when running
the system.
 The maximum controllable drilling parameters (e.g., WOB and RPM) that can be run for a given
formation or hole section.
 Lessons learned from offset wells including WOBs and RPMs that have worked well for the
relevant formations. These parameters can be used as initial starting locations when calibrating
the system.
 Plans for managing formation transitions: follow system recommendations or control drill (ROP
or DOC control) through transition zones.
 How often to calibrate the system.
SPE-184615-MS 11

Feedback from the field to the office is also important and can occur through a combination of remote
monitoring of the system via the mud logger’s web-based application and direct communication with
personnel at the well-site. Experience running the system in the field may help identify opportunities for
increasing (and sometimes decreasing) the controllable drilling parameters beyond (or below) the initial
“limits” communicated from the office. Additional “learnings” from the field can potentially feed into
the engineering redesign process (BHA, bit, drill string, motor, etc.) which can be applied to improve
drilling performance in subsequent wells. A flowchart summarizing the basic workflow of running the
surveillance and optimization platform and how the system is related to engineering planning is shown
in Figure 7.

Figure 6 – Various snapshots of the User Interface for the drilling surveillance and optimization platform as the system progresses
through calibration and eventually into optimization mode for a manufactured data example in WOB control. Progression is from
left to right and then top to bottom. Response points are small black dots and calibration points are larger dark green squares. The
performance heat maps shown in these images are for the normalized objective function and are colored green to red on a relative
scale where green is “better” and red is “worse”.
12 SPE-184615-MS

Figure 7 – Flowchart summarizing linkage between pre-drill, real-time drilling and post-drill workflows.

Results from Field Applications


In this section we present results from recent applications of running the surveillance and optimization
platform in the field. With each example, we emphasize workflows that are enabled by the system
which are able to drive gains in drilling performance.

Example #1
A lateral production hole section of a drilling operation targeting an unconventional oil and gas resource
was selected as an initial field application for the updated surveillance and optimization platform.
Although earlier versions of the system had been run previously in the field [see for example the field
results reported by (Chang et al. 2014) and (Payette et al. 2015)] this was the first field application of the
system using the updated User Interface described in this paper. This well was selected in part due to
the limited number of offsets in the area which created an opportunity for real-time drilling parameter
management to determine WOB and RPM “sweet spot(s)”. The lithology of the production section was
homogeneous with few formation transitions which limited the need for frequent system calibrations.
SPE-184615-MS 13

The system was utilized in the early stages (approximately first 1,500 m) of the production hole, just
past the landing point, using the same conventional bent-housing mud-motor assembly that had drilled
the build/curve section. Subject matter expects were onsite to provide training to operations personnel.
In this application the system was able to provide operations staff with the confidence to increase the
applied WOB by about 30% relative to existing practices for the area. This led to a significant increase
in ROP, a reduction in MSE and delayed the bit trip by about 1 day. Figure 8 shows traces of drilling
parameters for a 24 hour period during which the system was run. Figure 9 shows trends in the OBJ
surface following system calibration for two different snapshots in time, before and after identifying the
opportunity to increase the WOB limit from 14 kDaN (existing field practices) to 18 kDaN due to the
identification of efficient drilling conditions at the higher applied bit weights.

Figure 8 – Drilling parameter trends during a 24 hour period of using the system.
14 SPE-184615-MS

Figure 9 – Trends in the OBJ surface following two separate system calibrations. The left image shows the OBJ trend with 14 kDaN
set as the upper WOB limit and the right image shows the heat map trend following calibration with the WOB limit set to 18 kDaN.

Example #2
In a second example, the system was applied to assist in drilling an intermediate hole section in a
separate field, also targeting an unconventional oil and gas resource. The well had a typical L-shaped
profile with a 3,300 m vertical section and a 1,300 m lateral. The intermediate section had presented a
number of drilling challenges for the offsets in the area due to high stick-slip vibrations which were
largely associated with several hard layers of rock including one particularly hard 180 m thick
formation. As the offsets had required trips to replace the bit and/or downhole mud-motor above or
within the 180 m hard formation, surveillance strip charts were setup for monitoring the “status” of the
bit and motor to help identify when to trip (if necessary). A summary of some of the basic goals for the
application included:
 Rate of penetration: Maintain ROP at at least 10 m/hr.
 Vibrational dysfunction: Monitor MSE and TSE and maintain both as low as possible.
 Bit status: Monitor and maintain DOC ≥ 0.04 in/Rev. and μ ≥ 0.7.
 Motor status: Monitor Diff P and the Stand Pipe Pressure (SPP).
 Operation window: WOB between 8-18 kDaN, surface RPM between 30-50.
 System calibration: Calibrate surveillance and optimization platform once per stand (if
possible) and whenever lithology changes.
Figure 10 shows the OBJ response surface / heat map results following two separate instances of
system calibration. In the first case, see Figure 10(a), calibration was performed starting at low WOB
and RPM and then increasing parameters. In the second case shown in Figure 10(b) a different search
schema was adopted for calibration, starting first at high WOB and RPM and then varying the rotary
speed. In each case, following the calibration process the system was able to determine
recommendations for optimal set points for the controllable drilling parameters that best managed the
trade-offs between drilling fast and minimizing dysfunction (as characterized by MSE and TSE).
In addition to ROP optimization, basic surveillance charts were introduced onto one of the PCs in the
driller’s cabin to monitor various aspects of drilling. As shown in Figure 11, ROP was displayed in the
“Performance” chart and MSE and TSE were grouped into a separate chart labeled “Downhole
Dysfunction”. DOC and µ (bit aggressiveness) were viewed as “Bit Status” and differential pressure
and torque were used to monitor “Motor Status” which was used to respond to / avoid motor stalling
SPE-184615-MS 15

events. The surveillance arrangement of data channels was straight forward for the driller to track and
helped enable decisions related to controllable drilling parameter management and when to calibrate the
optimization and surveillance system.

(a) (b)
Figure 10 – Screenshots of the OBJ surfaces / heat maps following two separate instances of system calibration in the 180 m thick
hard formation. In each image the calibration points are numbered from oldest to most recent.

Figure 11 – Basic surveillance strip charts for monitoring drilling performance.

The combination of real-time controllable drilling parameter optimization along with bit and mud-
motor surveillance enabled the drill team to eliminate a planned trip and a new drilling record was
achieved for the intermediate hole section. Figure 12 compares the overall days for this hole section
against offset wells in the area with similar well paths and formations. Although the intermediate hole
section drilled using the system did not have the highest on-bottom ROP (as compared with offsets), the
total number of days required to drill the section was lowest due to the ability to finish the section with a
single bit run.
16 SPE-184615-MS

Figure 12 – A comparison of overall drilling performance for a number of offset intermediate hole sections and the test hole section
where the surveillance and optimization platform was run.

Example #3
In a third example the surveillance and optimization platform was applied repeatedly on one test drilling
rig and performance was compared to that of a neighboring offset rig that was not using the system. The
comparison was designed to minimize the influence of factors affecting drilling performance. Both rigs
were drilling intermediate hole sections through similar formations using nearly identical drill strings,
bottom hole assemblies, downhole mud-motors and (at the outset) similar practices for managing
controllable drilling parameters. The operator’s drilling practices were mature in this field with over one
hundred offset wells drilled prior to the study.
Comparisons were conducted between the two rigs using two different bit types with identical
frames. The only difference between the bit designs was the cutters; one bit used single chamfer and the
other dual chamfered cutters. Both rigs ran each of these bits and comparisons were made separately for
each bit. The system was used on the test rig to find combinations of controllable drilling parameters (in
this case WOB and surface RPM) for the various formation layers making up the intermediate hole
section that resulted in high ROP, efficient drilling (low MSE) and to the extent possible low levels of
stick-slip. Optimal controllable drilling parameters were determined through guided drill-off tests
commenced by user initiated calibrations of the system without the assistance of subject matter experts.
On the second rig, drilling personnel were provided guidance for managing controllable drilling
parameters based on experience from recent offset wells in the area.
Drilling hours vs. depth results from the two studies (single chamfer and dual chamfered bit designs)
are shown in Figure 13. On average, drilling proceeded about 2.5 hours faster per 1,000 ft when using
the surveillance and optimization platform. This gain was achieved primarily as a result of the system
encouraging the use of higher WOB (when appropriate) than was run in the offset wells. The dual
chamfered bit comparison in Figure 13(b) highlights the potential performance gains that can be
achieved as a result of system use. In this case, the platform was able to reduce the learning curve
associated with running the dual chamfered bit by quickly identifying efficient and improved drilling
conditions associated with running 10 to 15 klbs additional WOB as compared to the offset rig.
SPE-184615-MS 17

Figure 13 – The drilling hours vs. depth comparison demonstrates faster drilling with the optimization system for the intermediate
hole section for both: (a) the single chamfer bit and (b) the dual chamfered bit. IADC dull grades are also shown for each bit run.

A summary of normalized quarterly intermediate hole section drilling costs for rigs in the area (Q4
2014 through Q2 2016) and the test rig (Q3 2016) is shown in Figure 14. In this chart, all costs are
normalized using Q3 2016 vendor rates (bit rental and damage, drilling hourly costs, tripping hourly
costs) starting in Q4 2014 and ending in Q3 2016. The surveillance and optimization platform was
introduced into the drilling program for the test rig in Q3 2016 and resulted in a 40% cost reduction
relative to the previous two quarters. The system was credited with “direct savings” of 15% which were
attributed to real-time management of controllable drilling parameters through active use of the system
recommendations. Additional “indirect savings” were associated with a number of BHA redesigns
based on opportunities identified by the system to increase WOB (i.e., “design improvements”) and
changes to controllable drilling parameter management practices enacted apart from system usage (i.e.,
“drilling mechanics improvements”).

Figure 14 – Normalized quarterly drilling costs for the area for the intermediate hole section, with the surveillance and optimization
platform introduced into the drilling program on the test rig in Q3 2016. Note that quarterly costs are average normalized costs for
multiple rigs in the area with the exception of "Q3 Test Rig", which is the normalized quarterly cost for the test rig only.
18 SPE-184615-MS

Conclusions
The real-time surveillance and optimization platform was developed to enable better execution of the
ROP optimization process. The system leverages surface drilling data and provides vibration analysis
capabilities, advanced trending analysis tools, guided drill-off test workflows and various displays that
enable optimization of the drilling process. The current implementation of the system is advisory only,
i.e., no closed-loop control, with a focus on displaying pertinent information relating to drilling
performance and controllable drilling parameter management that best enables decisions and actions to
be taken by operations personnel at the well-site.
The system has been effectively used to provide value through real-time management of drilling
parameters, even when applied to drilling applications located in mature fields with many offsets. Post-
drill learnings obtained as a consequence of running the system have also enabled redesign of drilling
assemblies and drilling programs which have further led to enhancements of drilling performance.

Acknowledgments
The authors wish to thank the many field and office personnel who have supported the testing and
development of surveillance and optimization platform described in this paper. We would like to
particularly thank: Jeff Moss, Paul Pastusek and Tony Bautista from ExxonMobil; Trevor Holding,
Mark Mengers, Nathan Hilsendager, Daniel Go and Morgan O’Brien from XTO Energy; and Lars
Olesen, Behtash Charkhand, Edmund Quan, Terry Yee, Mariano Calvo, Vincent Dansereau, Sergey
Khromov, Steven Sheldon and Subodh Shrestha from Pason Systems. We are grateful to the
management of ExxonMobil, XTO Energy and Pason Systems for permission to publish this work.

References
WITS standard specification. Available from http://home.sprynet.com/~carob/.
Chang, Dar-Lon, Gregory S. Payette, Darren Pais, Lei Wang, Jeffrey R. Bailey, and Nicholas D.
Mitchell. 2014. Field Trial Results of a Drilling Advisory System. In International Petroleum
Technology Conference.
de Wardt, John P., Thomas B. Sheridan, and Amanda DiFiore. 2016. Human Systems Integration: Key
Enabler for Improved Driller Performance and Successful Automation Application. Society of
Petroleum Engineers.
Ertas, Deniz, Jeffrey R. Bailey, Lei Wang, and Paul E. Pastusek. 2014. "Drillstring Mechanics Model for
Surveillance, Root Cause Analysis, and Mitigation of Torsional Vibrations." doi:
10.2118/163420-PA.
Evangelatos, Georgios I., and Michael L. Payne. 2016. Advanced BHA-ROP Modeling Including
Neural Network Analysis of Drilling Performance Data. In IADC/SPE Drilling Conference and
Exhibition: Society of Petroleum Engineers.
Greenwood, Jeremy A. 2016. Vibration Monitoring and Mitigation - An Integrated Measurement
System. Society of Petroleum Engineers.
Jiang, Wanyi, and Robello Samuel. 2016. Optimization of Rate of Penetration in a Convoluted Drilling
Framework using Ant Colony Optimization. In IADC/SPE Drilling Conference and Exhibition.
Fort Worth, Texas, USA: Society of Petroleum Engineers.
Kristjansson, Sean D., Adam Neudfeldt, Stephen W. Lai, Julian Wang, and Dean Tremaine. 2016. Use
of Historic Data to Improve Drilling Efficiency: A Pattern Recognition Method and Trial
Results. Society of Petroleum Engineers.
Mirani, Ankit, and Robello Samuel. 2016. Mitigating Vibration Induced Drillstring Failures Using Data
Analytics: Workflow and Case Study. Society of Petroleum Engineers.
Nabavi, Josef, Rolv Rommetveit, Just Sverre Wessel, and Laurie Scott. 2016. A Challenging HPHT
Operation Supported by Dynamic Real Time Simulation, Forecasting and 3D Visualization.
Society of Petroleum Engineers.
SPE-184615-MS 19

Parasuraman, R., T. B. Sheridan, and C. D. Wickens. 2000. "A model for types and levels of human
interaction with automation." IEEE Transactions on Systems, Man, and Cybernetics - Part A:
Systems and Humans no. 30 (3):286-297. doi: 10.1109/3468.844354.
Payette, Gregory S., Darren Pais, Benjamin Spivey, Lei Wang, Jeffrey R. Bailey, Paul Pastusek, and
Michael Owens. 2015. Mitigating Drilling Dysfunction Using a Drilling Advisory System:
Results from Recent Field Applications. In International Petroleum Technology Conference.
Salminen, Kent, Curtis Cheatham, Mark Smith, and Khaydar Valiulin. 2016. Stuck Pipe Prediction
Using Automated Real-Time Modeling and Data Analysis. Society of Petroleum Engineers.
Schils, Saskia, Reina Teelken, Brian van Burkleo, Ove Johnny Rossa, and Nils Edwards. 2016. The Use
of Wired Drillpipe Technology in a Complex Drilling Environment Increased Drilling Efficiency
and Reduced Well Times. Society of Petroleum Engineers.
Teale, R. 1965. "The concept of specific energy in rock drilling." International Journal of Rock
Mechanics and Mining Sciences & Geomechanics Abstracts no. 2 (1):57-73. doi:
http://dx.doi.org/10.1016/0148-9062(65)90022-7.
Teelken, Reina, Colin Jaring, Andrew Craig, Roy Allan, Huw Roberts, Andy Hatch, and Nicol
Shepherd. 2016. Implementation of Wired Drill Pipe Saved Multiple Days Per Well, by
Addressing Performance Limiters, Increasing Drilling Efficiency. Society of Petroleum
Engineers.

Вам также может понравиться