You are on page 1of 7

Case Study: Understanding and Improving ESP Reliability in SAGD Wells

with High Dogleg Severity.


Alan Watt, John Graham, Alejandro Camacho, Sergey Belyaev, SPE
This paper was prepared for presentation at the 2013 Society of Petroleum Engineers Gulf Coast Section Electric Submersible Pump Workshop held in The Woodlands, Texas
April 24-26 2013.
This paper was selected for presentation by the ESP Workshop Panels (Rotating and
Permanent) following review of information contained in an abstract submitted by the
author(s). Contents of the paper, as presented, have not been reviewed by the ESP
Workshop Panels and are subject to correction by the author(s). The material, as
presented, does not necessarily reflect any position of the ESP Workshop or its panel
members. The author(s) retain copyright to this paper and have given permission to the
ESP Workshop to publish it in proceedings (electronic and hardcopy). Any other electronic
reproduction, distribution, or storage of any part of this paper for commercial purposes
without the written consent of the author(s) is prohibited.

Abstract
The steam assisted gravity drainage (SAGD) wells for an
operator in Northern Alberta often have dogleg severities
(DLS) between 6 to 14/100 ft, which are drastically higher
than established ESP recommended limits. Because of the
nature of these SAGD wells, the ESPs are also landed in a
nearly horizontal position. Typically, two to three
premature failures each year in the ESP wells have been
related to equipment bending. Inspection of the units
showed evidence of shaft breakage caused by bending
fatigue.
To understand the nature of these premature breakdowns,
the first step was to perform an investigation and determine
the main contributors with the help of stress calculation and
failure analysis. The calculations illustrated that any ESP
deployed horizontally has enough weight to follow even
extreme casing curvature, irrespective of ESP string outer
diameter (OD) or length. The investigation provided
evidence that a significant contributor to high bending
stress levels is ESP equipment OD changes. The second
step was to confirm the calculations by measuring the
deformation from bending at ambient conditions on a
typical ESP assembly. Equipment design changes, along
with improved practices of ESP placement and support in
the wells, have been implemented.

After the implementation of these practices, shaft breakages


due to bending have been eliminated and ESP runlife has
been extended. This work has been a significant contributor
of 9 to 10% year-on-year mean time to failure (MTTF)
increase.
The practices described in this paper will help engineers to
improve the completion design for ESPs in horizontal well
applications with greater than recommended dogleg
severity.
Introduction
After an evaluation of gas lift and ESP at the SAGD project
in North Alberta all production wells on Pads M and N
were converted to ESPs starting from June 2005. To handle
high total-fluid-flow rates and downhole temperature from
these wells, high-temperature 675-series pumps were
chosen. The pumps were driven by high-temperature
motors with advanced metal bellows protectors and bottom
feeder gas separators (BFGS). The ESP was landed in the
heel of the production well where the angle to vertical is
close to 90. In the SAGD process, an operator must
balance steam supply, production rates, and plant
operations in addition to considering ESP runlife. This
requires regular ESP speed changes which is realized by
using a variable speed drive (VSD).
The producing wells have the wellbore configuration as
seen on Fig. 1. The 13-in. intermediate casing is landed at
close to 90. The horizontal slotted liner has two blank
joints that overlap inside the intermediate casing. A debris
seal packer (DSP) is used as the liner hanger and seals off
inside the intermediate casing. A 5.5-in. tailpipe is run to
the toe of the producer with a DSP set in either the liner
overlap or the intermediate casing. The ESP is landed as
close to this DSP as practical to provide adequate fluid
levels for the ESP.

Fig.1Schematic of a typical SAGD production well,


mechanical lift.

Alan Watt, John Graham, Alejandro Camacho, Sergey Belyaev

History of the problem


The first shaft failure was identified in an ESP protector
that was pulled in March 2007 after 484 days of runlife.
During the dismantle inspection and failure analysis
(DIFA), it was determined that shaft failure was caused by
the failure of the thrust bearing. At that time, the 484 days
was considered a good runlife for the harsh SAGD
environment. See Table 1 for a summary of ESP failures
between 2007 and 2010 that were related to the shaft.

Usually when these indicators are observed by the operator,


the ESP would be stopped, and after some time, there
would be an attempt to restart the equipment. This is done
to ensure that the indicators above do not correspond to gas
locking of the pump or pumping the well off. If there is still
no flow to surface, an attempt will be made to pump some
liquid (usually water) through the tubing with ESP shutdown, while manually measuring the voltage, to see if there
is a backspin.
Initial analysis
The first step in the analysis was to isolate the failure cases
with less than 1 year runlives. Cases where the shaft failure
mechanism was not purely due to fatigue, but rather due to
extensive wear and heating of the shaft resulting from fluid
and sand ingress into the protector, have been excluded.
It is worth noting that this type of thorough analysis would
not be possible without the pictures and inspection reports
that were completed for all the failed units by the DIFA
group.

Table 2 in the addendum lists the ESP strings with shaft


failures after less than 1 year of runlife. All of the failed
ESPs consisted of a 675-series pump, LSMPMPM
Table 1Shaft failures history.
protector, either 256 or 147 horsepower 562-series motor
and BFGS, or an AGH with an integrated intake. Time
The wells were originally drilled under the premise that gas downhole ranged from 34 to 235 days. In most cases, the
lift would be used as the lift method; dogleg severities shaft failure occurred in the protector, but there were also
between 6 to 14/100 ft are common. These values are BFGS and AGH shaft failures.
drastically above established ESP operating recommended
limits. The influence was identified when DIFAs were As the motor usually keeps spinning after the shaft parting
completed on several ESP systems pulled in 2007. The occurs, the cross section of the broken shaft is subject to
runlives of those ESPs were short and only averaged about additional grinding that produces polished surfaces as seen
40 days, and industry-wise, this is considered a major in Fig. 2:
event. The location of the failure was the protector, BFGS,
or advanced gas handler (AGH).
Most times, the shaft failure was identified before the
equipment was pulled to surface. There are three main
indicators that can be used for this purpose. The first one is
the motor amperage. During normal operation, provided the
motor is loaded >40%, the amperage can be estimated from
ESP diagnostics software as a function of brake horsepower
of the pump. When the shaft parts between the intake and
the head of the motor, the amperage of the motor instantly
drops to the idle amperage, which is usually 30 to 40% of
motor nameplate amperage.
Another indicator is the loss of flow. If the choke is
automated and is set to maintain constant pressure, the
parting of the shaft will cause it to close as there is no fluid
being lifted by the pump.
The third indicator is the intake pressure that will rise after
the shaft is parted.

Fig. 2Broken protector shaft.

2013 ESP Workshop

Case Study: Understanding and Improving ESP Reliability in SAGD Wells with High Dogleg Severity.

In a fatigue mode of failure, the cyclic alternating loading


causes microscopic cracks to form at the stress
concentrators. Eventually a crack will reach a critical size,
and the shaft will suddenly fracture. In the cases described
here, the crack started in the splines area. Fig. 3.1 and Fig.
3.2 show examples of shaft failures.

Fig. 5---Schematics of waterfall curve.

Fig. 6Deformed ESP, a schematic of a worst-case scenario.

Visual modeling
To simulate the deformed shape of the ESP string, the 3D
models of the wells with shaft failures were created based
on the completion diagrams and well trajectories. It was
interesting to see that the waterfall curve in the ESP
landing zone took place in both the vertical and the
Fig. 3.1 and 3.2Broken shaft.
horizontal planes. Most of the wells modeled had a
Table 2 from the addendum, well trajectories, completion positive extremum in the projection to vertical plane. Fig.
diagrams, as well as the inspection reports and pictures 7.1 and Fig. 7.2 show a 3D model of the casing that was
were sent to Engineering Group for analysis. The group used to visualize the ESP landing section of the wellbore.
came back with the hypothesis that the shaft failure was
caused by the cantilever effect that took place when the
Vertical
ESP was landed at the setting depth, as illustrated in Fig. 4.
planes

Casing
Fig. 4Cantilever bend.

In the horizontal part of the wells the distributed weight of


components forces the equipment to conform to the
waterfall curve of the casing, as illustrated in Fig. 5.

Fig. 7.13D models of the casing to visualize the ESP landing


section.

Alan Watt, John Graham, Alejandro Camacho, Sergey Belyaev

Bending Experiments
How much will an ESP bend? To investigate this, several
experiments were performed in the service companys
shop. The first test was to confirm the stiffness of the
assembled ESP string. An ESP system was assembled and
placed on the floor to check the cantilever effect. The ESP
system was then blocked at both ends to simulate an 8/100
ft DLS. Additional blocks were then used to simulate
centralizers placed at several points to reduce the bending
of the system.

Casing cross
sections

Straight line reference

The test procedure is outlined in Procedure 1 of the


addendum.

For the first test the ESP system with 675-series pump was
raised at the ends until the center just came off the floor.
Fig. 7.23D models of the casing to visualize the ESP landing section. This provides the maximum amount the system will bend
when unconstrained at room temperature conditions. In the
heated wellbore the ESP should bend even more. From this
Engineering investigation
test (see Table 3) the average bend was calculated to be
A thorough investigation including stress calculations and equivalent to a DLS of 17/100 ft. The data was plotted in
failure analysis were performed to determine the main Fig. 9 showing the system centerline and the outline of the
contributors of these failures. Two important conclusions components outside diameters in Table 3.
can be drawn from the study. First, and contrary to
common conception, equipment OD does not play an
important role in ESP bending when horizontally deployed.
Any ESP deployed horizontally has enough weight to
follow even extreme casing curvature, irrespective of ESP
string OD or length. Second, a significant contributor to
high-bending stress levels is ESP equipment OD missmatch. This can cause a cantilever bend on adjacent
equipment, an effect that can only happen on horizontal
deployment. In the case studies, failures were located
between a 675-series pump and a 540-series protector. This
cantilever bend effect produced highly concentrated stress
in the location of the shaft failures.
Waterfall curve

Fig. 8Stress in the BFGS base and the protector shaft.

Table 3Bending of ESP string.

2013 ESP Workshop

Case Study: Understanding and Improving ESP Reliability in SAGD Wells with High Dogleg Severity.

Stabilizers
PROTECTOR

MOTOR

height above floor level (inches)

PUMP & INTAKE

Finally, an ESP system was blocked to simulate an 8/100


ft DLS. This system was then blocked to simulate
centralizers placed at several points to reduce the bending
of the system (see Fig. 10). The first system used a single
support. By trial and error the best location was 21.5 ft
from the pump end of the 44 ft system (see Fig. 11).

Fig. 9 Deflection of ESP.

PROTECTOR

MOTOR

Pseudo derivative (inches/foot 2)

PUMP & INTAKE

Fig. 11Deflection with a single support.

MOTOR

16

PROTECTOR

PUMP & INTAKE

height above floor level (inches)

Next, two calculations were performed to determine where


the bending was greatest. The first was the use of a best fit
curve to the centerline of the ESP system, as shown in Fig.
9 above. The centerline of the system drops below the best
fit line showing there is greater bending flexibility in the
protector and intake section of the ESP system than in the
motor and pump components. Secondly, the slope of the
bend was calculated for every 2-ft section along the system
and plotted as a pseudo derivative curve (see Fig. 10). This
shows that the greatest changes in slope were along the
intake and protector components confirming that these
components see the highest bending stresses.

14

12

10

Distance from pump support on right, (feet)

Fig. 12Deflection with a single support.

Distance from motor support on right (feet)

Fig. 10 Slope derivative.

For the second longer system two supports were placed


until the minimum bending occurred. On this 54 ft system,
the best locations were at 15 ft and 29 ft from the pump end
of the ESP (see Fig. 13 and Fig. 14 for details).

This concluded that for placement of ESP systems in a


deviated well with DLS, it is best to place the intake and
protector in the lowest DLS section available.
The ESP system was then placed on the floor (see Fig. 6)
and the motor section touched the floor 24 ft from the point
where the 6.75-in. pump crosses over to the intake.
Calculating this curvature yields a 1/100 ft DLS created by
the cantilever effect.

Fig. 13Deflection with two stabilizers.

Alan Watt, John Graham, Alejandro Camacho, Sergey Belyaev

16

MOTOR

PROTECTOR

PUMP & INTAKE

height above floor level (inches)

14

12

10

the easiest improvement was to move the ESPs to lower


DLS sections. Lower DLS sections were usually found
higher up in the wellbore, and special care had to be taken
to have enough fluid level above the ESP during
production. Fig. 16 shows the distribution of DLS.

Distance from pump support on right, (feet)

Fig. 14Deflection with two stabilizers.

Installation
A trial was planned for placing an ESP system in a well
with an 8/100 ft DLS. The well had a history of short run
lives because of bearing and shaft seal failures from the
high DLS it was landed in. The wells DLS was entirely
inclination rather than azimuthal such that centralizers
could be placed on the ESP system to reduce the bending.
The intermediate casing was 13.375- in. with a 12.4-in.
drift internal diameter. Stabilizers with vanes to allow
instrumentation and ESP cable to pass through were
designed with a 12-in. OD. These were welded to the top
of the protector and motor, in accordance with the best
placement of centralizers determined from the experiment
shown in Fig. 14. Based on this design, the maximum DLS
of the centerline would be dropped from 8/100 ft to 5/100
ft. The equipment was installed in July 2011 and pulled in
July 2012. From the DIFA, the shaft bearings and seals
showed much less wear than previous ESP systems pulled
with similar run life.

Fig. 16Distribution of DLS.

In some cases the lowest DLS was found to be across the


liner hanger (13 in. to 10 in.) or the crossover (10 in.
to 9 in.) in the liner overlap section. In these cases,
stabilizers with a varying diameter were used (see Fig. 17).

Stabilizers for changes in wellbore ID


Many wells were drilled with the configuration of 13.375in. intermediate casing with a liner hanger and 10.75-in.
blank crossing over to a 9.625-in. blank before the slotted
liner. A 5.5-in. tailpipe is landed just before the slots with a
debris seal packer. In many wells, the lowest DLS is in this
area with steps of IDs. In order to place an ESP system
across the step changes, stabilizers were designed to Fig. 17The set of stabilizers for liner overlap.
maintain a straight centerline of the ESP system (see Fig.
15 in the addendum). Stabilizers were placed about 10 to A requirement to leave a tangent section with low DLS in
14 ft apart to minimize the amount of bending between the last 200 ft before the slotted liner was communicated to
the drilling group. The recommendation was given to
them.
centralize the blank overlap section and to use two 10-in.
blanks to maintain a uniform bottom without steps (see
Decision making
Fig.18 in the addendum). The 200 ft tangent section should
In all cases the first priority was to have the intake and provide an acceptably small DLS in either the 13 in.
protector assembly in the lowest DLS, since this section is intermediate casing prior to the liner hanger (preferred
most susceptible to bending. The motor and pump sections location), the 10-in. blank section, or across the liner
can better withstand the higher DLS. For ESP placement, hanger with stepped diameter stabilizers (least preferred).

2013 ESP Workshop

Case Study: Understanding and Improving ESP Reliability in SAGD Wells with High Dogleg Severity.

When DLS took place predominantly in the inclination


plane, it was decided to run two stabilizers near the middle.
For example, in one of the wells the DLS of the ESP axis
was improved from 8 to 5/100 ft.
In other cases, where the ESP with a 675series pump
couldnt be landed higher up and stabilizers could not be
used due to the high azimuthal DLS, the 562-series pump
with 5.62-in. OD was used to conform to the OD of the
motor.
Conclusion
The study on the premature shaft failures has identified that
the main contributors to the failure were the cantilever
effect and the high DLS of the wells at the pump setting
depth. Good understanding of the deformed shape of the
landed ESP helped to come up with the decision-making
process for ESP replacement as well as planning the
trajectory and completion of the future wells. As a result,
shaft failures have been completely eliminated, which has
contributed to the increase in MTTF.

You might also like