Вы находитесь на странице: 1из 17

SPE/IADC-173052-MS

SMART Kick Detection; First Step on the Well Control Automation Journey
J. D. Brakel and B. A. Tarr, Shell; W. Cox, Noble Drilling Services Inc.; F. Jrgensen, and H. V. Straume,
National Oilwell Varco

Copyright 2015, SPE/IADC Drilling Conference and Exhibition


This paper was prepared for presentation at the SPE/IADC Drilling Conference and Exhibition held in London, United Kingdom, 1719 March 2015.
This paper was selected for presentation by an SPE/IADC 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 or the International Association of Drilling Contractors and are subject to
correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers or the International Association of Drilling
Contractors, 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 or the International Association of Drilling Contractors 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/IADC copyright.

Abstract
In 2010, based on the safety performance results achieved through the automation revolution in its oil &
gas downstream business, an operator set out to initiate the same revolution in its upstream business.
Automating the initial well control response to an influx was identified as the initial focus area with the
goal of assisting rig personnel to identify and stop any influx without delay. This led to a well control
automation collaboration project being initiated between an operator, a rig contractor and a rig equipment
supplier. The first phase of the project was to develop a system that could detect an influx across a broad
spectrum of well construction related rig operations. This paper describes the development, deployment,
and field testing of the first upgraded kick detection system from this collaboration.
To understand where to focus the kick detection system upgrade efforts, a fault tree style sensitivity
analysis of kick detection and well shut-in procedures was undertaken. The results pointed to the high
value of improved sensor data (both accuracy and reliability) and of improved detection software for
alarming (both in terms of coverage and how the driller is alerted to respond to a confirmed kick
condition). Based on this sensitivity analysis, a kick detection system upgrade functional specification was
created and used to develop a trial upgrade plan for a deepwater rig.
Initial implementation operational performance results are presented to demonstrate that most of the
kick detection system upgrade objectives were achieved.
Operational feedback from using the upgraded kick detection system is included which highlights the
new, SMART, features that were designed to provide easily understood alerts to the driller; including
unique pop-up kick alarm windows for drilling or circulating, making a connection, and tripping in or out.

Introduction
Kick detection is a critical activity during well construction operations that typically involves the Driller
monitoring various inputs to determine if there is any indication of an influx of formation fluids into the
well. These inputs include a combination of crew member reporting of visual observations and sensor data
displayed on the Drillers console. Through extensive well control training, Drillers learn to detect a kick
(or unwanted influx of formation fluid into the well) by comparing actual values with the planned, or

SPE/IADC-173052-MS

expected, normal values for each well construction operation conducted after installing the subsea
blow-out preventer equipment on the well.
From a kick detection perspective, the majority of well construction operations can be categorized into
one of the following three activities that require the Driller to apply different kick detection methods:
1. Drilling or circulating
2. Making connections
3. Tripping in or out.
It is extremely important to respond to any positive kick indicator as quickly as possible to minimize
the size and consequence of the influx that is allowed to enter the wellbore before the well is secured. Well
control certification training programs and on-the-job kick-drills provide Drillers with training to apply
the appropriate kick detection methods to these three activities. The primary kick indicators for drilling
or circulating are increase in mud return flow rate (vs. mud flow in) and pit gain. For making connections
continued return flow with pumps off and pit gain are the primary indicators. For tripping in or out
(including being out of the hole and flow checking), continued return flow and trip tank gain (vs. expected
change) are the primary indicators. However, vendors of drilling rig control systems have been slow to
realize the desirability of automated kick detection systems that can alert the Driller to a possible kick
situation, regardless of the well construction activity being conducted.
Given the challenge to find a way to automate the initial well control response to a kick, it soon became
apparent that setting a goal to provide the Driller with a reliable kick detection system that uses the three
different kick detection methods, as currently taught for each of the above listed activities, would be a
good starting point.
To better understand what a SMART kick detection system might look like for a deep-water floating
rig, the essential kick detection and well shut-in response steps were identified for each of the relevant
three well construction activities listed above. By combining these steps with appropriate reliability data
it was possible to conduct a sensitivity analysis on a number of the key variables to help define the
opportunities for designing reliability into any new kick detection system.
One of the key challenges that many well control projects face is finding representative statistical data
for well kicks. Two of the most important variables needed for this project were overall kick frequency
per rig-year, and the relative frequency of kicks detected during the most relevant well construction
activities. Industry statistics provided an overall kick frequency but the relative frequency of kicks was
more difficult to determine as current industry practice is to report kicks detected during connections in
the same category as Drilling Kicks. A brief review of historical deep-water floating rig kick data, from
the operator and rig contractor parties in the project, resulted in the following relative kick frequency
being used in the analysis:
25% Drilling or circulating
25% Making connections
50% Tripping in or out.
check list above - numeric data was missing
Note that these values are not necessarily representative of industry experience but are reported here
to illustrate that, based on the experience of one operator and drilling contractor, half the historical
deep-water floating rig kicks reported as Drilling Kicks were actually detected when making a
connection. More recently (Fraser 2014) estimated the percentage of connection related kick events to be
even higher at 70% of all kick events.

SPE/IADC-173052-MS

Kick Detection Fault Tree Analysis


To understand the value of possible improvements to current kick detection systems in common use on
deep-water floating rigs, the kick detection and well shut-in fault tree analysis included evaluation of the
following sensitivities:
1. Base
2. Base
3. Base
4. Base

case
case Best Sensors
case Best Sensors Smart Alarms
case Best Sensors Smart Alarms Automated Shut-in.

The Base case fault tree was built to represent the typical performance of a Driller deployed on a
modern deep-water floating rig using the kick detection equipment supplied with the rig. Performance was
defined as the sum of the initial influx detection performance, the well shut-in decision making
performance and the well shut-in performance.
The Base case Best Sensors fault tree provided the opportunity to envision the potential incremental
performance improvement that could be achieved by replacing existing sensors with the highest accuracy
kick detection sensors available and including a duplicate set of sensors (using different base technologies) that could then provide robust health checking of all kick detection variables.
The Base case Best Sensors Smart Alarms fault tree provided the opportunity to envision the
potential incremental performance improvement that could be achieved by adding kick detection software
capable of alerting the driller to a kick as early as possible with both a unique audible alarm and an easy
to understand, operationally relevant, pop-up kick alarm window.
The Base case Best Sensors Smart Alarms Automated Shut-in provided the opportunity to
envision the potential incremental performance improvement that could be achieved by completely
automating the well shut-in process.
The results of the fault tree sensitivity analysis are reported in Table 1 in terms of the performance
improvement vs. the Base case. From Table 1 it is clear that the biggest improvement in kick detection
and shut-in performance (16.9%) was predicted to be from including Best Sensors. Adding Smart Alarms
resulted in a smaller than predicted, but still significant, additional performance improvement of 9.1%.
Perhaps the most surprising result of the analysis was that the predicted incremental performance
improvement from adding Automated Shut-in was only 1.3%. However, such may be attributable to the
Drillers high degree of kick detection and well shut-in competency assumed in the Base Case analysis.

Table 1Kick Detection and Shut-in Performance Fault Tree Sensitivity Analysis Results
Sensitivity Case

Total Improvement

Incremental Improvement

Base case Best Sensors


Base case Best Sensors Smart Alarms
Base case Best Sensors Smart Alarms Automated Shut-In

16.9%
26.0%
27.3%

16.9%
9.1%
1.3%

As a result of the fault tree sensitivity analysis, a project was initiated to upgrade the kick detection
system on a deep-water floating rig with two focus areas:
1. Using industry recognized effective kick detection sensors and duplicating all sensors
2. Developing new kick detection software that included operationally specific alarms.

SPE/IADC-173052-MS

Functional Specification for SMART Kick Detection


Objectives
Based on the sensitivity analysis, it was clear that the proposed SMART kick detection system upgrade
for the targeted deepwater floating rig had to include the following features:
1. Best practical sensor technology with duplicate measurements for all key parameters using (where
possible) different sensor types
2. Robust rig operations event detection to identify which kick detection technique was applicable
3. Specific kick detection algorithms for the different rig operation events when different kick
detection techniques are required
4. Distinctively different audible kick alarm from other drilling system alerts and alarms
5. Easily understood kick alarm pop-up window display for each type of kick detection technique.
Per one industry standard (NORSOK Standard D-001 2012), redundancy of pit volume, flow return
and standpipe pressure sensors are already recommended. Earlier revisions also recommended different
measuring principles, except for standpipe pressure, were there is no good alternative measuring principle.
Per one industry safety system guideline applicable offshore Norway (Norwegian Oil and Gas
Association 2004), current kick detection systems are not required to have a Safety Integrity Level (SIL)
rating. However, one of the key objectives for the functional specification for the proposed SMART kick
detection system was to incorporate relevant instrumented safety system features into the functional
specification so that a SIL rating could be established in preparation for potential future certification as
an instrumented safety system.
In line with the required features and the safety system objective, the proposed SMART kick detection
system functional specification was divided into the following work scopes:
a. Sensor Health Checking to trigger ALERT condition if general comparison of independent
calibrated sensor outputs measuring the same parameter indicates a significant difference (e.g.
more than X % maximum difference based on expected sensor accuracy, repeatability and
dead-band performance). Duplicate sensors (preferably different sensor types) expected to be
provided for Flow-in (from any selection of rig pumps and/or cementing unit pumps), Stand Pipe
Pressure (rig floor), Flow-out, Active & Reserve pit levels (where any number of the reserve pits
can be designated as part of the Active System with the primary active pits) and Individual Trip
Tank level measurements (A, B and stripping tank). For pit levels need minimum of one sensor
pair, e.g. mounted in opposite corners, to compensate for floating rig motions. Hence, need
minimum of two pairs for pit level measurement duplication on floating rigs.
b. Sensor Data Consistency Checking to trigger ALERT condition if general correspondence of
sensor data deviates from recent finger printing (e.g. within maximum Y % of expected).
Appropriate sensor response norms during drilling (or circulating), connections (or flow checking)
and tripping in/out events will form the basis of this finger printing to give early identification
of a possible kick.
c. Robust Event Detection logic to enable kick detection system to be aware of which kick alarm
triggers and pop-up screen needs to be active corresponding to drilling (or circulating), connections (or pumps off) or tripping (in/out or flow checking). Note that for kick detection, total
pumping rate includes all pumps contributing to the return flow in the riser.
d. Sensor Selection Checking to trigger ALERT condition if any required sensor data is missing
due to sensor failure or failure to select the correct set of sensors (or missing input value required
for kick detection) based on automated detection of rig status (e.g. drilling, connection or tripping).

SPE/IADC-173052-MS

e. Apply modern noise reduction and pattern recognition signal processing for earlier and more
informative kick detection alarming (audible kick ALARM and kick alarm pop-up screen). This
was expected to provide the opportunity of setting tighter (and more robust) kick detection
thresholds than currently possible using noisy raw measured data. Apply pattern recognition to the
analysis of the flow back at pumps stop signatures to provide the basis for improved detection on
connections.
Additional Sensors
A rig survey was conducted to verify which sensors were already installed and to find suitable locations
for the required sensors to provide full redundancy of key measurements. The survey looked at additional
pit volume sensors, standpipe pressure sensors, return flow sensors and stroke sensors for all pumps.
Redundant sensors for standpipe pressure were already in place and a spare slot for redundant stroke
sensors was available on all pumps.
The possibility for installing a Coriolis flowmeter for measuring mass flow, density and temperature
on the return line was investigated but rejected due to the smaller than needed change in elevation of the
return flowline. The rig already had a paddle-type flow sensor on the return line and the onboard mud
logging company had added a radar flow sensor. Since there was no available space to add another flow
sensor, sharing the output from the mud loggers radar flow sensor provided the needed duplication of
return flow measurement.
Accurate pit volume data is an important factor in kick detection, but the pitch and roll of floating rigs
impacts the accuracy of pit level measurements. Instead of just duplicating the existing pair of level
sensors in each of the active and reserve pits, some thought was given to also improving overall accuracy
of the resulting pit volume data. To better eliminate pitch and roll effects, the four level sensors planned
for each pit were combined in sensor pairs and mounted in opposite corners diagonally across from each
other; one pair with guided microwave sensors and the other with non-contacting radar sensors.
One of the challenges with fitting all the pits with four sensors is that the pits were not designed for
optimal volume measurements but space efficiency. The level measurement sensors must be placed at a
minimum separation distance from mud guns, walls, pipes and agitators inside the pit. At the same time,
the location of piping and walkways on top of the pit, where the sensor housing and cables are mounted,
also needs to be taken into consideration. Overall this did not allow placement of the sensors exactly
where they were initially planned, but still sufficiently close to each corner for each pair to be effective,
as shown in Figure 1.

SPE/IADC-173052-MS

Figure 1Optimized locations of mud pit level sensors based on rig survey

New Software
Since the pop-up Kick Monitoring Display concept had already been developed by the rig equipment
supplier in 2011, based on functional design requirements for the type of information that would help the
drilling team recognize a potential kick, the concept was ready to be integrated into this project. The
pop-up displays were designed to not only reinforce the audible kick alarms but also provide enough alarm
validation data to guide the Driller in regards to confidently making the decision to shut-in the well (or
not). Figure 2 shows an example of the Kick Monitoring Display pop-up activated by a drilling or
connection related kick alarm.

SPE/IADC-173052-MS

Figure 2Example pop-up window for drilling (pumps on) or connection (pumps off) influx events

The objective of this pop-up display (for drilling and making connections) was to reduce the amount
of information which must be actively monitored while grouping the information in a manner which
allows the crew to take actions appropriate for the kick being experienced. For example, for connections
it gives the crew a tool to finger print the current connection by comparing it to return flow signatures
of the last five connections.
For drilling and making connections three key kick indicators (Active Pit Volume Gain/Loss,
Return Flow Increase while drilling and Return Flow with zero SPM during connections) are used to
trigger the distinctive audible kick alarm and the associated pop-up Kick Monitoring Display. The kick
alarm sound for these three alarms is a dedicated second stage alarm tone that is distinguishable from all
other alarms in the Drilling Instrumentation & monitoring System onboard.
To ensure the kick alarm pop-up does not prevent the driller from monitoring essential operational
parameters, the pop-up is configured to not overlay any of the key parameters on the real-time drilling and
tripping display on the Drillers console.
The Active Pit Volume Gain/Loss alarm notifies the Driller if the pit volume gains or losses exceed
pre-set parameters and was already a part of original drilling instrumentation scope. The change in regard
of Pit Volume Gain/Loss High alarm was to enable the second stage audible alarm tone for this alarm
to be distinguishable and to avoid unnecessary alarms during connections, as shown in Figure 3.

SPE/IADC-173052-MS

Figure 3Explanation of Active Pit Volume Gain/Loss with Alarm Suppression during Connections

The Return Flow Increase while drilling alarm notifies the Driller when the return flow is increasing
more than expected while drilling. It requires the Driller to set limit values for SPM, Time and Return
Flow. (See Figure 4.) In order to identify when the flow is increasing at an unacceptable level, the system
needs a return flow reference point. This is done automatically or manually. The automatic calculation is
done in the following steps that are illustrated in Figure 4:

Figure 4 Explanation of Return Flow Increase while drilling alarm

1. If the Active SPM is different from the Active SPM set point ( SPM Limit) then the Return
Flow Increase while drilling alarm is disabled.
2. When the Active SPM is equal to the new Active SPM set point ( SPM Limit) the Time
Limit time counter starts.
3. The Time Limit time counter ends and the flow is expected to be stable at this point in time. The
Average Time then starts and the system begins to collect the return flow data.
4. The Average Time ends and the new Average Flow is calculated. The Return Flow Increase
while drilling alarm is then enabled.
5. If the return flow exceeds Average Flow Flow Limit the alarm is triggered.

SPE/IADC-173052-MS

Finally, the Return Flow with Zero SPM alarm notifies the Driller when there is an unacceptable
level of flow that continues while the pumps are turned off. It requires a time input defined by the Driller
and a Flow (%) alarm value. As an example, the Driller can set the alarm to trigger if the Flow out level
is not below 5% within XX seconds after the Active SPM has been ramped down to zero. The alarm will
be enabled at the end of the configured time span from the point where Active SPM falls under 3 SPM.
(See Figure 5.)

Figure 5Explanation of Return Flow with Zero SPM alarming

In this project, kick alarming was expanded to include a tripping operations specific kick alarm and
corresponding informative pop-up Kick Monitoring Display, consistent with the drilling and connection
kick alarming design and usability described above. The tripping kick indicator used was Abnormal trip
tank difference Gain/Loss, where abnormal is defined as varying from the theoretical steel displacement
volume for that point in the tripping operation, as shown in Figure 6.

Figure 6 Explanation of Abnormal trip tank difference Gain/Loss alarm

The Drilling instrumentation system already had a trip tank totalizer difference calculation installed
that is designed to compare the change in the total trip tank totalizer volume in real-time with the amount

10

SPE/IADC-173052-MS

of pipe tripped in or out (displaced). Therefore the scope was to add audible alarms to the existing
calculations and develop the corresponding pop-up display. Figure 7 shows the pop-up display that was
developed to provide an explanation of any kick alarm triggered during tripping.

Figure 7Example of the pop-up window display triggered during tripping

Once the Driller sets the parameters and configures the information on the screen as he would like to
see it, the system should require no additional maintenance/adjustments during normal operation. The
system continually monitors the well, on behalf of the Driller, while only requiring his engagement if
alerted of a potential situation or if he desires to adjust parameters. The parameters are set in the same
application and in the same manner as the rest of the audible alarms in the drilling instrumentation system.
Whats important to note is that only the Driller has the ability to configure and adjust kick alarm
parameters, yet the pop-up Kick Monitoring Display is visible on all the workstations located on the rig,
including the office areas, so that multiple crew members are alerted in the event of a kick alarm.

Kick Detection System Upgrade


Installation and Commissioning of Additional Sensors
As commissioning was to be conducted in a small time window between wells, the new remote I/O cabinet
required and all the additional sensor cables were installed prior to commissioning, leaving only the new
pit level sensors to be installed during commissioning.
Prior to installation of the additional pit level sensors, detailed sensor installation and commissioning
instructions were issued, together with sensor user manuals containing detailed information on mounting
restrictions, with regards of pipes, walls, agitators and mud guns. To locate and calibrate the new sensors,
the pits containing mud had to be emptied and cleaned.
After connecting to the radar return flow sensor (via a splitter), an attempt was made to linearization
the output to counteract for low response. Despite best efforts, a good correlation with the paddle flow
sensor data could not be obtained. Therefore, the paddle flow sensor was selected as the primary return
flow sensor and the radar flow sensor was selected as the back-up return flow sensor.

SPE/IADC-173052-MS

11

Installation and Commissioning of Software


Following acceptance testing in Norway, the SMART kick detection system was installed and commissioned on the rig during a time period that was between wells and in conjunction with a training module.
The training module consisted of a class room walkthrough and a live demonstration of all the new
functionality and configuration features. After commissioning, onboard training continued when the
system was operational on the next well.
The software installation and commissioning sequence was as follows:
1. Backup all the existing software and computers
2. Install new software on all remote workstations in the accommodations
3. Install new server to host new software
4. Install new network router to accommodate the additional components in the system
5. Install new software on the drilling controls servers and programmable logic controller
6. Final commissioning of the system.
The full backup was performed to provide a rapid restore option in case of any issues during the
installation of the upgrade. Upgrading of the remote workstations was done first as these are not essential
for operations on drill floor. Installation of the new software on the drilling control servers and
programmable logic controller was done while the equipment was not being used by the drill crew; i.e.
upgrading of the Driller server was done while the drill crew was using the Assistant Driller server and
vice-versa. The key to avoiding any system downtime was good communication. Final commissioning
and sign off on the system was completed in one day to minimize the associated non-productive time for
the drill crew. Keeping the drill crew informed every step of the way allowed them to plan ahead and
integrate each installation step seamlessly with other planned between well rig activities.
Rig Crew Familiarization and Training
How does one introduce a completely innovative idea and/or tool to a veteran drill crew? The intent of
the new SMART kick detection system was to improve early kick detection and ultimately minimize both
the time taken for the system to detect and initiate a kick alarm and the time the driller needs to confirm
the validity of the alarm and begin to shut-in the well. In order for the system to function as intended,
drilling personnel must be trained to navigate the system, input correct data, and recognize/acknowledge
any warnings or alarms. At the time of commissioning the only place available for training was on the rig.
Finding time when training could be conducted without interfering with operations was an issue, as
hands on training was desirable. In order to facilitate hands-on training of the drill crews, a field
engineer was sent to the rig to assist as necessary. While onboard, the field engineer shadowed the Driller
and Assistant Driller in the drill shack to help navigate through the system. The field engineer taught
personnel how to set all drilling parameters and alarm limits, how to recognize alarm banners/audible
horns; and also advise personnel to investigate any alarm accordingly. Having a field engineer onboard
for an extended period (21 days) to work with multiple crews of drilling personnel was beneficial.
However, for future installations, much of this hands-on training can be achieved using an onshore
training simulator before the system is installed offshore. To that end, plans have been initiated to create
a rig specific course that incorporates kick detection alarms, banner alerts, and screen displays. This
course can act as the main training opportunity for all rig personnel and would be held in a rig specific
simulator. The intent being that the course would support their Well Control training requirements and
would be routinely retaken on a periodic basis (e.g. every other year as a refresher for all necessary rig
personnel). Should there be any new updates to the SMART kick detection system, an updated training
course would provide the opportunity for participants to learn those updates before seeing them on the rig.

12

SPE/IADC-173052-MS

Implementation Results
Rig Crew Operational Feedback
From the inception of the system, rig personnel were kept informed to prevent any surprises during
implementation. After the system was installed and commissioned, focus shifted from the developers to
the end users. Rig personnel served as the main source of feedback, e.g. what worked well and what did
not work so well. The implementation process was a two-way conversation between the drill crew users
and the system developers. As the drill crew continued with operations, any abnormalities experienced
were recorded and discussed on a bi-weekly basis with all parties involved. System developers responded
to end-user feedback and updated the system as necessary.
As previously mentioned, one added feature of the system was the installation of extra pit sensors to
minimize natural rig motions such as pitch and roll but what about motions created by deck cranes?
Before the system was installed any crane usage had to be suspended once operations dictated that all pit
level fluctuations be closely monitored to prevent any false or inaccurate readings. Rig personnel
discovered that the added sensors not only neutralized normal rig motions but also permitted free use of
all cranes without interference to accurate pit volume monitoring.
Influx Detection Performance
The best way to speak about the newly installed SMART kick detection system performance to date is to
describe its response during an actual small influx event that occurred a few days after the new kick
detection system was installed and commissioned.
In that instance, the on-tour Driller experienced a well influx while making a connection. The driller
began to decrease the mud pumps strokes to zero in preparation for another connection. A flow check was
then conducted by the drill crew. From previous connections, the average pumps-off time for return flow
to decrease to zero had been 90 seconds. The Driller waited the normal 90 seconds and then asked a
member of the drill crew to perform a visual flow check. The crew member reported continuous flow and
the Driller reacted by shutting-in the well. A few minutes later the Driller successfully completed the well
shut-in operation.
Figure 8 identifies the recorded sequence of events, as shown by the time based flow-in (pump strokes)
and flow-out (radar flow sensor) recorded data, and includes annotations related to when positive influx
indicators were seen. Note that the first positive indication of an influx occurred just seconds before the
pumps were shut-down, based on the small detected increase in return flow that triggered the new, kick
specific, audible alarm and the corresponding new pop-up window. However, since the Driller was ready
to make a connection, the Driller simply acknowledged the Return Flow Increase while drilling alarm
and proceeded with the planned flow check that was part of the routine connection procedure he was
using.

SPE/IADC-173052-MS

13

Figure 8 Time based recorded data showing continuing return flow after pump shut-down

Before the Driller completed the well shut-in operation, increasing return flow triggered the Return
Flow with Zero SPM positive influx indicator, resulting in the second kick specific audible alarm and
corresponding pop-up window event, as illustrated in Figure 9. Note that in Figure 9 it can also be seen
that the observed small return flow rate, after pump shut-down, was not easy to detect from the paddle
style return flow measurement (selected for this plot) but was later identified from the positive slope seen
in the total pit volume response, after the initial pump shut-down transient. No alarm was triggered by the
abnormal pit volume response (or finger printing), after pump shut-down, as this type of positive influx
indicator was not included in this first generation SMART kick detection system. However, a second
generation SMART kick detection system is now under development that could include an abnormal
finger printing type positive influx alarm that will be active when the mud pumps are shut-down.

14

SPE/IADC-173052-MS

Figure 9 Annotated time based data plotted on the Drilling System for the influx event.

False Alarm Rate Performance


Since alarm thresholds were historically set by the Driller, based on the expected well conditions and
general detection performance of the rigs kick detection system, the decision was made early in the
project to use the same approach for the SMART kick detection system. This decision meant that the
actual in-service false alarm rate performance of the SMART kick detection system would be linked to
the various manually set influx detection thresholds.
However, since the SMART kick detection system was designed to use different detection methods for
drilling, making connections and tripping it was anticipated that all the potential false alarms that can now
be triggered on current generation systems by transient events, such as starting and stopping pumps, could
be significantly reduced.
To establish the actual false alarm rate performance of the SMART kick detection system all the alarms
designed to trigger the unique audible kick alarm and trigger any of the Kick Monitor Display pop-ups
were monitored during the first six months after the installation and commissioning activities were
completed. The results of the monitoring are summarized in Table 2, False Alarm Frequency of the
SMART Kick Detection System.

SPE/IADC-173052-MS

15

Table 2False Alarm Frequency of the SMART Kick Detection System


Alarm Variable

False Alarm Frequency (%)

Kick Indicators
Return Flow Increase (Drilling)
Return Flow with Zero SPM (Connection)
Active Pit Volume Gain (Drilling or Connection)
Trip Tank Volume Difference Gain (Tripping)
Total % Kick Alarms

0
2
31
2
35

Loss Indicators
Return Flow Decrease (Drilling)
Active Pit Volume Loss (Drilling or Connection)
Trip Tank Volume Difference Loss (Tripping)
Total % Loss Alarms

36
28
1
65

Table 2 shows that strong in-service false alarm rate performance was achieved for three of the four
kick indicators with a false alarm rate of zero for Return Flow Increase (and a 100% detection rate based
on the single influx event detected), and 2% each for both Return Flow with Zero SPM and Trip Tank
Volume Difference Gain. However, the false alarm rate for the fourth kick indicator Active Pit Volume
Gain was unacceptably high. Investigating this high false alarm rate revealed that the Active Pit Volume
Gain threshold is set based on early kick detection during drilling and the alarm was being triggered at
each connection when the normal flow back volume to the pits (e.g. more than 25 bbls) exceeded the much
smaller threshold value (e.g. 5 10 bbls) typically used for kick detection while drilling ahead. At first
it was thought that increasing the pit volume gain threshold value would eliminate these false kick
alarms but doing so would have compromised early kick detection while drilling.
Table 2 also shows that the largest proportion of false alarms (65%) were related to two of the three
loss indicators that also trigger the distinctive audio kick alarm (but do not trigger the pop-up kick
monitoring display). Fortunately both the Return Flow Decrease and Active Pit Volume Loss
indicators only triggered a false alarm when the rig pumps were stopped and started respectively, e.g. at
the start and end of making a connection.
Although there was a high frequency of false alarms, they did not occur randomly but seemed to be
associated with starting and stopping the pumps. Hence, they did not detract from the overall value of the
enhanced kick detection provided by the SMART kick detection system.
The rig equipment supplier used this performance data to identify improvements in the alarm threshold
setting methodology for Pit Volume Gain and identify ways to deal with the observed false alarming
when pumps are started or stopped. These improvements are intended to be incorporated into the next
generation SMART kick detection system now under development.

Conclusions
1. An examination of available deep-water floating rig kick data from the operator and drilling
contractor, who agreed to collaborate with a rig equipment supplier to develop an improved kick
detection system, revealed that approximately half the kicks historically reported as drilling kicks
actually occurred when pumps were shut down to make a connection.
2. From a detailed fault tree style analysis, a SMART kick detection system that includes at least two
independent sets of high quality sensors, preferably using different sensor technology, to measure
all key kick detection input values was predicted to provide the largest improvement of kick
detection and well shut-in performance.
3. A trial, automated, kick detection system that mimics the three different kick detection methods
used by competent rig crews for kick detection during drilling, making connections and tripping

16

SPE/IADC-173052-MS

has been implemented in a SMART kick detection system developed collaboratively by an


operator, rig contractor and rig equipment supplier.
4. The combination of a distinctive audio kick alarm and an associated informative pop-up display,
configured to provide relevant information to aid the Driller in making a prompt well shut-in
decision, have been built into this SMART kick detection system.
5. Implementing the SMART kick detection system on an existing deepwater floating rig required a
comprehensive rig survey to identify the new locations (and the cabling etc.) required for the
additional flow out, flow in and pit level (active, reserve and trip tanks) sensors. Final installation
of the additional sensors and commissioning of the new software was completed between well
operations (e.g. when the subsea blow-out preventer was on deck and all the pits were empty/
accessible). Due to effective planning and coordination, the upgrade work was completed without
incurring non-productive rig time.
6. Due to the collaboration between the operator, rig contractor and rig equipment supplier, the
SMART kick detection system was welcomed by the rig team, and efforts expended to implement
it effectively through appropriate classroom style and hands-on training.
7. Shortly after commissioning the SMART kick detection system, a small influx was detected during
a connection that validated the effectiveness of the new connection kick detection methodology
built into the system. Post-analysis of the event confirmed that the first kick alarm (and the
associated informative pop-up display) was a drilling kick alarm that occurred, and was acknowledged by the Driller, a few seconds before the pumps were shut down to make the connection. A
second kick alarm was triggered a few minutes later during the connection flow check, after the
pumps were shut down, and the well was shut-in.
8. Increased crew confidence in the SMART kick detection system was evident after the influx event.
However, the improved stability of pit volume data also gave the rig team the opportunity to
continue using rig cranes without concern of compromising kick detection when using a small
Active Pit Volume Gain threshold value.
9. Analysis of the in-service false alarm rate over a period of 6 months indicated excellent
performance with respect to three of the four kick alarm indicators. However, the Active Pit
Volume Gain threshold value that proved very effective for early kick detection when drilling
ahead resulted in a false kick alarm whenever the pumps were shut down and the flow back volume
to the active pits exceeded the Active Pit Volume Gain threshold value. Two additional false
alarms associated with loss detection indicators were also being routinely triggered when pumps
were started or stopped. (Improvements needed to eliminate these nuisance alarms have been
identified and are planned to be implemented in the next generation SMART kick detection
system.)
10. The reported overall success of the SMART kick detection system field trial on a deepwater
floating rig can be attributed to the strong collaborative relationship that was established between
the operator, rig contractor and rig equipment supplier. All parties contributed to the success of the
project by sharing in the common goal to develop a SMART kick detection system to support
faster decision making; using kick detection practices, which experienced rig crews are already
familiar with, for drilling, making connections and tripping.

Acknowledgements
Special thanks must first go to Shell, Noble Drilling and National Oilwell Varco for permission to publish
this paper. The authors would like to acknowledge the strong management support given to this project
in our respective companies and recognize all the project team members for their valuable assistance.
Special thanks for their invaluable contributions go to the offshore rig crew members who ensured the

SPE/IADC-173052-MS

17

successful implementation and smooth transition to routine daily use of the new SMART kick detection
system on their rig.

References
1. Fraser, D. et alet al. 2014 Early Kick Detection Methods and Technologies. Paper SPE 170756MS, presented at the SPE annual Technical Conference and Exhibition, Amsterdam, 27-29
October.
2. NORSOK Standard D-001 Drilling Facilities, 2012 (Annex B), Norwegian Technology Standards
Institution, Oscarsgt. 20, Postbox 7072 Majorstua, N-0306 Oslo, Norway.
3. Norwegian Oil and Gas Association 070, Application of IEC 61508 and IEC 61511 in the
Norwegian Petroleum Industry, Rev 02, 29 Oct 2004.

Вам также может понравиться