Open Access

Journal of Energy and Power Technology

Original Research

A Data-Based Reliability Analysis of ESP Failures in Oil Production Wells

Shaikha Al-Ballam 1, Hamidreza Karami 2, * Deepak Devegowda 2

1. Kuwait Oil Company, Ahmadi, Kuwait; E-Mail: [email protected]

2. Norman, Oklahoma, United States; E-Mails: [email protected]; [email protected]

* Correspondence: Hamidreza Karami; E-Mail: [email protected]

Academic Editor: Francesco Gabriele Galizia

Special Issue: Energy Efficiency in Flexible and Reconfigurable Manufacturing: Emerging Trends,
Models and Applications in the Industry 4.0 Era

Journal of Energy and Power Technology Received: September 27, 2022

2022, volume 4, issue 4 Accepted: November 02, 2022
doi:10.21926/jept.2204036 Published: November 14, 2022

Abstract
Electrical Submersible Pumps (ESPs) are one of the most widely used artificial lift methods in
the petroleum industry. However, ESP failures are unanticipated and common occurrences
with significant financial impacts for the operators. Analysis of the ESP performance and
failures are essential in its design and optimization. This paper presents a statistical approach
for diagnosing and evaluating the root causes of ESP failures. The analysis is based on the field
data gathered from the surface and downhole ESP monitoring equipment over five years of
production of 10 wells. Electrical failures are the most common general cause of ESP failures,
accounting for 61% of all failures, followed by motor failure and gas locking. Specifically,
power failure, under-voltage, voltage unbalance, and motor underload are the most common
occurrences. The data trends are analyzed for the two weeks before each specific failure, and
conclusions are made on the warning signs to predict failures. In addition, a Weibull statistical
analysis model is constructed to evaluate the reliability features and estimate the ESP failure
probability, allowing operators to perform preventive maintenance. The results provide
guidelines for ESP operations and contribute to reducing or preventing ESP downtimes and
operating costs.

© 2022 by the author. This is an open access article distributed under the
conditions of the Creative Commons by Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium or format,
provided the original work is correctly cited.
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Keywords
ESP statistical analysis; ESP reliability analysis; field data; ESP failures; case studies

1. Introduction

With the ever-declining production rates from oil and gas fields worldwide, the demand for
efficient artificial lift techniques to enhance production is increasing daily [1]. The selection of a
suitable artificial lift technique for a given well is a function of various operational parameters [2].
However, the goal is always to maximize the profit from the well by increasing production while
maintaining the artificial lift-related expenses and downtime to a minimum. This is what makes a
thorough understanding of the operating ranges, limitations, and failures of an artificial lift
technique vital before its application.
Electrical Submersible Pumps (ESPs) are one of the most widely used artificial lift techniques in
the petroleum industry, especially for highly productive oil wells [3]. They can provide noticeable
increases in a well’s production if looked after and maintained within their optimum operating range.
However, ESP failures are usually sudden and unanticipated in the oil field and could become
common occurrences. These failures cause significant financial impacts for the operators due to oil
production deferments and high intervention costs [4]. In addition, excessive shutdowns and trips
significantly reduce the run-life of an ESP [5].
Production monitoring and surveillance of ESP wells help extend their run life. Furthermore, an
ESP monitoring system is highly beneficial for recording a pump’s performance and acquiring
valuable data on the downhole conditions. These data are used to analyze and predict failures
before they occur. These data involve various operations and field conditions, including dynamic,
static, and historical data. They are the key factors for data-driven models predicting ESP system
failures [6].
Mohrbacher and Tabe discussed ESP installation, operation, maintenance, and issues, including
a system failure analysis [7, 8]. From the 1990s through 2010, several authors used statistical models
to examine ESP problems. There were comparisons of failure distributions among the system
components [9], different equipment types and models [10], or different companies, platforms, and
fields [11]. In addition, historical trends have been used to assess the evolution of ESP run life
through time. Past studies [10-12] presented statistical distributions fitted to historical data to
forecast future failure frequencies. Sawaryn et al. emphasized the need to include all the ESP system
parts in the study [13]. They suggested that the simulation time be extended to encompass the
whole life of a field and appropriately measure ESP reliability. Furthermore, Alhanati et al. provided
a standardized ESP failure analysis that included a list of all ESP-related failure types, failed items,
failure descriptors, and root causes [14].
The failure rates of ESPs across an oilfield have been demonstrated to vary significantly [15].
Given the large number of ESPs installed in each oilfield, ESP failure prediction research mainly
concentrates on computing population-level estimations and failure causes [16]. Sawaryn
developed analytic terms to describe failure patterns at a population level [17]. Liu et al. also
modified the data mining classification techniques [18].

Page 2/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

1.1 Evolution of ESP Monitoring System Over Time

The monitoring levels of ESP systems have changed over time. The first and most straightforward
method of diagnosing an ESP system’s failure is using ammeter charts to measure and record the
current drawn by the downhole motor [4]. The wellhead pressure is then utilized to calculate the
pump's head in the wellbore, followed by fluid shots to measure the fluid level. Fluid shots are useful
but may be inaccurate owing to factors like foamy crude or completion configuration [19]. Recently,
it has been possible to safeguard an ESP by forcing a pump shutdown depending on underload and
overload current values, which is a sign of poor operating conditions.
The variable speed drive (VSD) was the next stage in the growth of ESP technology. This
equipment gives operators an extra means of controlling an ESP’s functions while protecting pumps
and motors from electrical stress. The main advantage of a VSD is that it allows the operator to
modify the ESP's speed in response to variations in the well's productivity index or changing well
conditions [20].
The introduction of the ESP Downhole Monitoring System was the next significant step in the
evolution of ESP monitoring and control. This system gathers and transmits data from the downhole
system to the surface, where it may be recorded and researched to optimize system performance
and longevity. These devices have helped to improve the accuracy of the most crucial production
metrics. They are becoming more prevalent, with a growing number of new installations equipped
with them [5].
The ESP Downhole Monitoring System analyses well characteristics and delivers pump data to
improve ESP efficiency and reserve recovery rates [21]. It helps to keep track of key operational
parameters, as indicated in Table 1.

Table 1 Main Parameters Recorded in an ESP Downhole Monitoring System [21, 22].

Parameter To determine:
Pump Intake Pressure (PIP), psi Static and flowing bottom hole pressures
Pump Intake Temperature (Ti), °F Gas Volume Fraction at the pump intake
Pump Discharge Pressure (Pd), psi Head and efficiency of the pump
Pump Discharge Temp (Td), °F Optional
Motor Temperature (Tm), °F Operating temperature rise (Tm – Ti)
Current Leakage (mA) Indication of impending ground fault conditions
Bearing mechanical wear and frequency that causes
Vibration (Vx & Vy)
excessive vibration (Resonance).
Motor Frequency, Hz Motor speed
Motor Current (I), Amps Changes in fluid density and power consumption

The aforementioned parameters indicate the most relevant aspects to consider when analyzing
ESP failures [23-25]. Even though appropriate monitoring helps delay system failures, they may still
happen for various reasons [19]. Because the electrical system is typically the weakest link in an ESP
system, most failures are electrical, resulting from a mechanical problem as the underlying cause of

Page 3/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

the failure [26]. Each failure must be thoroughly probed, and its fundamental reason must be
determined. Mubarak et al. [27] researched ESP failures in the Wafra field over four years and
discovered that the most common causes of ESP failures include motor failures (40%), pump failures
(22%), cable failure (26%), and others (12%). Furthermore, according to Al-Sadah [28], the most
common reasons for ESP failures based on the Dismantle, Inspection, and Failure Analysis (DIFA)
results are burnt motor (28%), damaged cable/MLE (27%), and damaged penetrator (24%). This
study uses a comprehensive set of data from multiple oil wells with ESP systems to further analyze
the ESP failures over time and explain the root physical causes.
The following sections detail the evaluation of statistical failure data, case studies of ESP failures,
and an overview of ESP-specific failures. This overview highlights the reported ESP-specific failure
modes and their corresponding early warning indicators in the ESP parameters, which may be used
as a general guideline to detect ESP operating difficulties more efficiently and with less effort. The
discussion then shifts to ESP reliability analysis, where correct assessment and forecasting of ESP
failure probability led to a greater understanding of system performance. Finally, the conclusion
outlines the most significant findings from the study.

2. Statistical Analysis Approach

This study presents a detailed analysis based on actual field data from Kuwait Oil Company (KOC)
assets. For the evaluation, these data are categorized into static, dynamic, and historical data from
over five years of production of 10 active wells. The dynamic high-frequency data was obtained from
surface and downhole ESP monitoring equipment (VSD, pump, and wellhead). The static data
includes the current well completion, ESP pump design, and reservoir fluid information. The dataset
incorporates the failure’s cause, the downtime’s duration, and the corresponding pump data.
Furthermore, historical operational data is used to supplement the analysis.
In our study, each well is equipped with gas separators to separate the free gas from the
producing fluid before entering the pump. In addition, the ESP pump motors used in our study are
typically two-pole, three-phase induction motors with squirrel cages. The motors operate at a
nominal speed of 3500 rpm at 60 Hz or 2915 rpm at 50 Hz with a two-pole configuration. They run
on three-phase power between 230 and 5,000 volts and between 12 and 200 amps. The diameter
and length of the motors determine their horsepower ratings, and since there are no wires running
along their lengths, the motors may be built somewhat larger than the pumps.

2.1 ESP Statistical Failure Data Evaluation

A statistical evaluation of the observed ESP failures in the available field data of this study is
investigated in depth. The findings of our analysis are depicted in Figure 1, dividing the failures into
a few main categories. In addition, Figure 2 shows a more specific division of the most statistically
important causes for ESP failures. The evaluation includes 971 ESP failures from over 5 years of
production from 10 wells. The detailed Specific Failures are shown in Table S1, Table S2, Table S3,
and Table S4. Some of the most witnessed general failure roots are:
 Electrical Failure (61%): includes power fail, under/over voltage, phase unbalance, and
overcurrent.
 Motor Failure (18%): includes motor voltage unbalance, motor overload, overheating, etc.

Page 4/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

 Gas Effect (13%): excessive gas content is locked in the pump, causing the current to drop
rapidly because of the motor underload.
General Descriptor
2% 1%

5%
Electrical Failure
13% Motor
Gas Effect
Reservoir Related
Power Cable
18% 61% Others
Mechanical Failure
External Influence

Figure 1 ESP Failure Statistical Analysis-General Descriptor.

Figure 2 ESP Failure Statistical Analysis-Specific Failure Mode (SFM).

2.2 Case Studies

In this study, two specific cases will be presented in detail to discuss the types of failures and
their corresponding effects on the ESP and the well’s production. The two most prevalent reasons
shown above were observed in the cases analyzed. The ESP in Well #1 experienced an electrical
failure, specifically phase unbalance. Well #2, on the other hand, experienced electrical failure
induced by an overcurrent. Both cases share a common general reason for failure but have distinct
underlying causes of failure. This highlights the need for recognizing and diagnosing the
fundamental cause of ESP failure to choose the most effective course of action to mitigate it. Each
case includes the failed ESP's history, the Dismantle Inspection, and Failure Analysis (DIFA)
outcomes, and the underlying cause of failure.

Page 5/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

2.2.1 Well #1

A downhole ESP pump is being used to produce oil from this vertical oil well with API gravity of
29°, average liquid production of 2,200 BPD, 76% water cut, and a gas-oil ratio (GOR) of 545 SCF/bbl.
This ESP was installed and run in the well for 290 days (9 months). Figure 3 depicts the wellbore
schematic for Well #1. The system suffered unstable intake pressure (Pi) and fluctuations in all
running parameters (wellhead pressure, flowline pressure, Vibration, Amps, etc.). The downhole
sensor readings were lost a week before the failure, and the system tripped on overload. The
preliminary cause of failure for this ESP was diagnosed as electrical failure resulting in a burn in the
Motor Lead Extension (MLE), as seen in Figure 4.

Figure 3 Well #1 Wellbore Schematic.

During the DIFA, the housing was dismantled, severe wear was found on all stages, a scale-like
material on the outside of the diffuser, and emulsion/foamy fluid on the outside of the top 16
diffusers, and a hole on the diffuser due to severe wear. Regarding electronics, the MLE cable was
blown out 24 ft above the pothead, and the main power cable had a thick layer of scale-like material.
Apart from the MLE cable, the cable was verified to be electronically sound.
Many factors contributed to the wear of the pump’s stages over its lifespan, including fluid type,
scale/solids, backpressure, and down-thrust. The formation's scale-like material caused severe wear,
resulting in power overload. Overload increased current draw, resulting in overheating and the
cable blowing out. This wear caused a hole in the pump, and as a result, the MLE cable at the hole
location was shorted. Table 2 summarizes the failure analysis results for Well #1 based on the DIFA
report. The root cause of the failure in this situation was the well’s Reservoir Performance.

Page 6/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Table 2 Well #1 Failure Analysis Clarifications.

Reason for Pull – General: Electrical

Reason for Pull – Specific: Phase Unbalance
Primary Failed Item: ESP Cable
Secondary Failed Item: Motor Lead Extension
General Failed Descriptor: Electrical
Specific Failed Descriptor: Short Circuit
General Failure Cause: Reservoir or Fluids
Specific Failure Cause: Reservoir Performance

A workover was performed to selectively re-perforate the formation based on Pulsed Neutron
Capture (PNC) log results for water saturation and hydrocarbon layers. The well was then completed
with a new ESP.

2.2.2 Well #2

A downhole ESP pump is being used to produce oil from a vertical oil well with API gravity of 30°,
average liquid production of 3,000 BPD, 60% water cut, and a gas-oil ratio (GOR) of 428 SCF/bbl.
This well was run with an ESP for 102 days (3 months). Figure 4 depicts the wellbore schematic for
Well #2.

Figure 4 Well #2 Wellbore Schematic.

The well faced electrical failure due to overcurrent. The motor had two melted leads and was
grounded. The cable suffered mechanical damage during the POOH, and its upper side had less

Page 7/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

insulation resistance. The impellers and sleeves of the pump showed radial wear, with some fallback
debris on the upper side. In addition, the pump stage washers had light upthrust wear.
Table 3 summarizes the Failure Analysis results for Well #2, based on the DIFA report. The most
likely root cause of failure was filtered clogging with foreign material, resulting in reduced flow along
the items inside the shroud and poor cooling. There was an overheat warning on the housing,
indicating that the Motor had overheated, resulting in electronic failure. As a result, an upper sand
trap was advised for installation.

Table 3 Well #2 Failure Analysis Clarifications.

Reason for Pull – General: Electrical

Reason for Pull – Specific: Overcurrent
Primary Failed Item: ESP Motor
Secondary Failed Item: Motor End Connectors (Y-point/Leads)
General Failed Descriptor: Material
Specific Failed Descriptor: Melted/Damaged
General Failure Cause: Reservoir/Fluids
Specific Failure Cause: Sand/Mud

2.3 ESP Specific Failures Overview

In this study, several distinct ESP failures are analyzed regarding the early warning signs before
the actual failures. The trends in the data are depicted with time, starting from two weeks before
each specific failure mode (SFM). Common warning signs are assessed using ten samples from each
SFM presented in this section. The objective is to have a general understanding of the data trends
associated with each SFM.

2.3.1 SFMs Associated with Electrical Failures

ESP electrical failures are prevalent and attributed to surface and downhole systems [14]. The
downhole system fails if any of the electrical components in the ESP assembly fails, including the
power cable, motor (i.e., stator), or downhole sensor. As shown earlier, electrical failures account
for 61% of all ESP operational failures. The trends in the acquired data before some of the ESP
electrical failures are shown in this section.

Under Voltage. The under-voltage specific failure is caused by the main power supply with
inadequate control or a wiring problem. This SFM can cause the pump motor to overheat, resulting
in motor failure. Multiple ESP variables are explored to identify the patterns before the actual failure.
Figure 5 depicts motor voltage and current data for two weeks before under-voltage failures for two
examples in Wells 2 and 4 of our datasets. The x-axis indicates the time before failure (TBF) in days.
Both voltage and current diverge from their typical trends at almost the same time in both wells.
The voltage of the motor in Well 2 begins to drop 0.8 days before failure and continues to decline
until the actual failure. For Well 4, a reduction in voltage is first seen 2.1 days before failure, while
the motor current increases. As the voltage of the motor drops, the motor draws greater current,

Page 8/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

which causes the motor to overheat, resulting in sudden failure. These wells display early warning
indicators of a condition requiring immediate attention.

Figure 5 Under-voltage Specific Failure Effects on Motor Current and Voltage.

Power Fail. A power fail occurs due to unbalanced phases, voltage spikes, the presence of
harmonics (distorted current and voltage), or lightning strikes. Consequently, the ESP motor and
power cable are overheated. Takacs recommended a clean electric supply or a VSD unit with
sinusoidal output to alleviate these issues [19].
Figure 6 depicts examples of motor voltage, motor current, and wellhead pressure for two weeks
before two power failures in Wells 3 and 6. The figure shows the declining trend of wellhead
pressure within a day before failure for both wells. In addition, voltage spikes (surges) are seen in
Well 3 beginning 0.78 days before failure. Similar trends are found for the Well 6 case, where the
spikes start 2.9 days before the failure. Voltage surge is the rapid increase in voltage over a very
brief period in a power system [4]. In addition, the motor current in Well 3 exhibits an increasing
trend three days before failure. A similar observation is made for Well 6, starting 2.8 days before
the failure, showing early warning signs of a problem that demands attention.

Page 9/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Figure 6 Power Fail Specific Failure Effects on Motor Voltage, Current, and Wellhead
Pressure.

2.3.2 SFMs Associated with Motor Failures

Most ESP motor failures are due to electrical issues [14]. Based on the investigated field data,
motor overload and high motor winding temperature are some motor-related failures. As shown
earlier, motor failures account for 18% of all ESP operational failures. The trends in the acquired
data before some of the ESP motor failures are shown in this section.

Motor Overload. An overload failure happens when the motor draws excessive current, resulting
in excessive power consumption. As a result, the motor may overheat, leading to motor damage
[14]. Various factors may contribute to an overload, including an improperly sized motor, an
unexpectedly high fluid specific gravity raising the Total Dynamic Head (TDH) over the design value,
and inconsistent motor voltage [15]. Figure 7 depicts wellhead pressure, motor current, and motor
temperature trends for the two weeks before two cases of overload failures in Wells 3 and 10. The
x-axis indicates the time before failure (TBF) in days. The figure shows the declining trend of
wellhead pressure as failure time approaches for both wells. Moreover, the motor current in Well
3 suddenly increases three days before failure. The trend is similar for Well 10, where the deviation
begins 1.8 days before the failure. The motor temperature also rises as the failure approaches,
beginning 2.7 days in advance for Well 10 and 2.1 days before for Well 3. These wells may exhibit
early warning signs of a problem that demands attention.

Page 10/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Figure 7 Motor Overload Specific Failure.

High Motor Temperature. A high motor temperature happens when excessive voltage supply or
drawn motor current leads to overheating issues [4]. Overheating is a primary cause of motor failure,
particularly when the motor is forced to work harder or is placed under an unexpected load. As the
speed of the motor and, by extension, the rotation of the pump's shaft rises, the moving
components get overheated due to the increased friction [29]. Figure 8 displays motor voltage,
current, and temperature before two cases of motor temperature failure in Wells 8 and 1. The x-
axis represents the TBF in days. The graph depicts the increasing trend of motor current as the
failure time for both wells’ approaches. In addition, the motor voltage in Well 8 surges rapidly 0.6
days before the failure. A similar observation is made for Well 1, where the deviation starts one day
before the failure. As the failure nears, the motor temperature increases, starting 0.8 days before
failures in both wells. These wells display early warning signals of a serious situation.

Page 11/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Figure 8 High Motor Temperature Specific Failure.

2.3.3 SFMs Associated with Other Failures

Motor Underload. An underload failure often indicates pumped-off or gas-locked conditions [19].
Different high-frequency ESP variables were investigated to observe the patterns before the failure.
Figure 9 depicts pump intake pressure and motor current during the two weeks before two
examples of underload failures in Well 8 and Well 9. The x-axis indicates the time before failure
(TBF). The figure shows the rising trend of pump intake pressure as failure time approaches for both
wells. Moreover, the motor current in Well 8 deviates from its regular pattern at 6.3 days, with a
decreasing trend as the failure approaches. A similar observation is made for Well 9, where the
deviation begins 1.8 days before the failure.

Figure 9 Underload Failure Prior Data Trends.

2.4 Summary

Page 12/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Table 4 summarizes the described specific ESP failure modes with their early warning signs in the
ESP parameters. The average time before the failure (TBF), when each sign starts to appear, is also
included based on the ESP failure data of this study. These qualitative observations can be used as
a general guideline to identify the ESP operational issues earlier and with less effort.

Table 4 Each ESP SFM with its Early Warning Signs, Based on the Data of This Study.

SFM Early Signs Average TBF (Days)

Motor current increases 1.1
High Motor Temperature Motor voltage increase 1.7
Motor temperature increases 0.8
Motor voltage increases 1.1
High Cable Temperature
Motor current increase 1
Wellhead pressure drops 3.1
Overload Motor current increase 2.4
Motor temperature increase 2.4
Plugged with sand Hard start followed by motor current spikes 2.5
Voltage spikes 1.85
Power Fail Current increases 2.9
Wellhead pressure drops 0.7
Sensor Failure Motor Voltage spikes and continue increasing 1.5
Motor voltage drops 3.7
Under Voltage
Motor current Increases 1.4
Pump intake pressure increases 2.5
Underload
Motor current decreases 3.9

3. ESP Reliability Analysis

This section discusses the reliability analysis of ESP failures, where the probability of ESP failure
is a key metric. Reliability analysis is an extensively studied statistical technique for forecasting
systems’ performance in various sectors [30]. It denotes the probability of fully functioning during
a specific time interval [30]. Mean Time Between Failures (MTBF) is strongly associated with this
concept, with a longer MTBF indicating greater system reliability [31]. Accurate evaluation and
forecasting of ESP failure probability contribute to a deeper comprehension of system performance.
Mean time between failures (MTBF) quantifies the average time it takes for a system to fail. MTBF
measurements may be used to evaluate ESP performance, design, and reliability. MTBF is defined
as the arithmetic mean value of the reliability function, R(t), which is the predicted value of the
time till failure density function, f(t) [31].
∞ ∞
𝑀𝑇𝐵𝐹 = ∫ 𝑅(𝑡)𝑑𝑡 = ∫ 𝑡 𝑓(𝑡) 𝑑𝑡 𝟏
0 0

𝑓(𝑡) = 𝜆𝑒 −𝜆𝑡 𝟐

where 𝜆 is the failure rate.

Page 13/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

For the case of this study, Equation 3 may be used to calculate the MTBF. In addition, Equation 4
may be used to calculate the failure rate.

∑(𝑆𝑡𝑎𝑟𝑡 𝑜𝑓 𝐷𝑜𝑤𝑛𝑡𝑖𝑚𝑒 𝑜𝑓 𝐶𝑢𝑟𝑟𝑒𝑛𝑡 𝐹𝑎𝑖𝑙𝑢𝑟𝑒 – 𝑆𝑡𝑎𝑟𝑡 𝑜𝑓 𝑈𝑝𝑡𝑖𝑚𝑒 𝑎𝑓𝑡𝑒𝑟 𝑙𝑎𝑠𝑡 𝐹𝑎𝑖𝑙𝑢𝑟𝑒)

𝑀𝑇𝐵𝐹 = 𝟑
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑎𝑖𝑙𝑢𝑟𝑒𝑠

1
𝐹𝑎𝑖𝑙𝑢𝑟𝑒𝑅𝑎𝑡𝑒 = ∗ 100 𝟒
𝑀𝑇𝐵𝐹

The ESP’s MTBF trend for the 10 wells over the five years is shown in Figure 10. Initially, all wells
had a maximum MTBF of 46 to 354 days. The MTBF decreases with time, particularly in the third
year of operation (2019), until it reached its lowest value in 2021. In 2017, Well 1 had the longest
MTBF of 354.5 days, while the lowest was 46 days for Well 6. After five years, in 2021, Well 8 had
the longest MTBF of 38 days, while Well 3 had the shortest duration of 10.9 days. The failure rates
of ESPs in the ten wells can also be calculated using Equation 4 throughout the five-year production
period. The failure rates of Well 6 and Well 7 are consistently more significant than those of the
other wells.

Figure 10 ESP’s MTBF for all wells over five years of production.

Now let’s focus on wells 6 and 7, as the wells with the highest failure rates. Figure 11 shows the
specific failure modes by subcategory for these wells. For Well 6, power failure accounts for 71% of
all SFMs, while the remaining 29% is almost evenly split across MTR cable, overload, under voltage,
and underload. For Well 7, power failure accounts for 72% of all SFMs, while overload is 15%,
Undervoltage is 9%, and overvoltage is 4%.

Page 14/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Figure 11 ESP’s Specific Failure Mode (SFM) for Well 6 and Well 7.

3.1 Weibull Analysis

Failure data can be modelled using various distributions, including normal, exponential, Rayleigh,
Weibull, gamma, and lognormal. Weibull Analysis is an efficient statistical data analysis technique
for assessing time dependence. The two-parameter Weibull distribution is defined as follows [15]:

𝛽 𝑡 𝛽−1 −(𝑥⁄𝜂)𝛽
𝑓(𝑡) = ( ) 𝑒 𝟓
𝜂 𝜂

where 𝛽 is the shape parameter, 𝜂 is the scale parameter, and t represents the value to be
evaluated.
The shape factor (𝛽) determines the distribution's behavior. The failure rate drops with time if
𝛽 < 1, meaning that the ESP gets more trustworthy as it matures. Manufacturing or installation
errors may be a cause of this. The failure rate increases if 𝛽 > 1, mainly due to pump wear. Finally,
if 𝛽 = 1, it denotes a consistent failure rate that is time independent.
Based on the field data from 10 wells, a Weibull model is built to estimate the probability of
failure. The previously estimated MTBF values are utilized as inputs to the Weibull analysis model.
Next, the Weibull shape and scale parameters are estimated using Median Rank Regression. Then
the Weibull Probability plots are generated. On logarithmic scales, these plots depict unreliability,
defined as the probability of failure (%) vs. time to failure (days).
Figure 12 illustrates Well 6's and Well 7's unreliability plots, respectively. The legend indicates
the number of MTBF points (5 points for five years), Weibull parameter estimates (β and η),
confidence limits used (Beta-Binomial Bounds), and the degree of confidence (90%). The Weibull
line accurately fits the MTBF points in both plots with R2 values of 99.4% and 86.5%, respectively,
and a low p-value. The predicted Weibull parameters for Well 6 are β of 1.538 and η of 28. A 𝛽 > 1
value is a sign of pump wear and increasing failure rate with time. According to the plot, the ESP in
Well 6 has a 90% probability of failure after 75 days of the previous failure. The probability of failure
increases to 99% in 80 days.

Page 15/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

Figure 12 Weibull Analysis for Failure Probabbilties of Well 6 and Well 7.

From Figure 12, the predicted Weibull parameters for Well 7 are β of 0.835 and η of 52.87, a
possible indicator of manufacturer errors in the early life of the well. According to the plot, the ESP
has a 90% probability of failure after 190 days. The predicted failure probability increases to 99% in
300 days. Comparing both ESP wells, well 7 is more reliable than well 6, as the 90 percent probability
of failure occurs after 190 days vs. 75 days in well 6.

4. Conclusions

This study analyzes the field data collected from surface and downhole ESP monitoring
equipment of 10 wells over a 5-year production period. According to these data, three common
categories of ESP failures are electrical failures (61%), motor failures (18%), and gas effects (13%).
Looking more specifically, power failure, under-voltage, voltage unbalance, and motor underload
are the most common occurrences. Two field case studies from two wells in Kuwait Oil Company
(KOC) assets are discussed in depth to explore the failure types and their related consequences on
the ESP and the well's production. These cases of electrical failure have distinct underlying causes
of phase unbalance and overcurrent, respectively. Diagnosing the root cause of ESP failure is
essential for identifying the best line of action to mitigate it.
The early warning signs of failures are investigated by looking at the data trends within the two
weeks before each SFM. On average, the warning signals of ESP failure manifest themselves two
days before the failure, depending on the SFM. The most impacted variables are the motor current
and voltage, consistent with the fact that most ESP failures are electrical for the data under
investigation.
The Weibull statistical analysis is used to assess the ESP reliability, predicting the probability of
ESP failure and the MTBF values. The ESP's MTBF trends for the ten wells are estimated over five
years. Initially, all wells had long MTBF values (higher reliability). The MTBF drops with time, starting
from the third year of operation. Consequently, ESP loses its reliability over time as operational

Page 16/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

issues cause an increasing number of failures. This model aids in the forecasting of ESP failure
probability.

Author Contributions

Conceptualization: H Karami and D Devegowda; Field data acquisition and literature review: S.
AlBallam and H Karami; Reliability analysis: S Alballam; Supervision: H Karami and D Devegowda. All
authors have read and agreed to the published version of the manuscript.

Competing Interests

All Authors wish to confirm that there are no known conflicts of interest associated with this
publication, and there has been no significant financial support for this work that could have
influenced its outcome.

Additional Materials

The following additional materials are uploaded at the page of this paper.

1. Table S1: The Likely Specific Causes of Electrical Failures.

2. Table S2: The Likely Specific Causes of Motor Failures.
3. Table S3: The Likely Specific Causes of Gas Effect Failure.
4. Table S4: The Descriptions of the Other ESP Specific Failures.

References

1. Igwilo KC, Okoro EE, Ubanatu S. Comparative approach to optimum selection of artificial lift
system. Pet Coal. 2018; 60: 429-437.
2. Nguyen T. Artificial lift selection methodology for vertical and horizontal wells in conventional
and unconventional reservoirs. In: Artificial lift methods: Design, practices, and applications.
Cham: Springer; 2020. pp. 317-347.
3. Fakher S, Khlaifat A, Hossain ME, Nameer H. Rigorous review of electrical submersible pump
failure mechanisms and their mitigation measures. J Petrol Explor Prod Technol. 2021; 11: 3799-
3814.
4. Takacs G. Electrical submersible pumps manual: Design, operations, and maintenance.
Netherlands: Elsevier Science; 2009.
5. El Gindy M, Abdelmotaal H, Botros K, Ginawi I, Sayed E, Edris T. Monitoring & Surveillance
improve ESP operation and reduce workover frequency. Proceedings of Abu Dhabi
International Petroleum Exhibition and Conference; 2015 November 9-12; Abu Dhabi, UAE.
Richardson, Texas: OnePetro. doi: 10.2118/177926-MS.
6. Abdelaziz M, Lastra R, Xiao JJ. ESP data analytics: Predicting failures for improved production
performance. Proceedings of Abu Dhabi International Petroleum Exhibition & Conference; 2017
November 13-16; Abu Dhabi, UAE. Richardson, Texas: OnePetro. doi: 10.2118/188513-MS.
7. Mohrbacher JD. A field study of ESP performance in a deep, hot, and sour environment.
Proceedings of SPE Rocky Mountain Regional Meeting; 1984 May 21-23; Casper, Wyoming, USA.
Richardson, Texas: OnePetro. doi: 10.2118/12913-MS.

Page 17/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

8. Tabe FL. An overview of the installation, operation, maintenance, and problems associated with
electrical submersible pump system in central Sumatra, Indonesia. Proceedings of SPE Annual
Technical Conference and Exhibition; 1984 September 16-19; Houston, Texas, United States.
Richardson, Texas: OnePetro. doi: 10.2118/13201-MS.
9. Higgs G. ESP performance – A statistical review. Proceedings of 3rd European Electrical
Submersible Pump Round Table; 1994; Aberdeen, Scotland, United Kingdom. BP Exploration.
10. Venkataraman G, Mikus T. Reliability analysis of electrical submersible pumps. Proceedings of
1994 SPE ESP Workshop; 1994; Houston, Texas, USA. Shell Development Company.
11. Oliveira LF, Bardy MB, Filho SS, Neves EA, Silva JA, Agustoni JA, et al. Analysis of ESP failure data
from the northeastern pole of the Campos oil basin. Proceedings of ESP Workshop; 1997 April
30-May 2; Houston, Texas, USA.
12. Patterson MM. A model for estimating the life of electrical submersible pumps. SPE Prod Fac.
1993; 8: 247-250.
13. Sawaryn SJ, Norrell KS, Whelehan OP. The analysis and prediction of electric-submersible-pump
failures in the Milne point field, Alaska. Proceedings of SPE Annual Technical Conference and
Exhibition; 1999 October 3-6; Houston, Texas, USA. Richardson, Texas: OnePetro. doi:
10.2118/56663-MS.
14. Alhanati FJS, Solanki SC, Zahacy TA. ESP failures: Can we talk the same language? Proceedings
of SPE Gulf Coast Section Electric Submersible Pump Workshop; 2001 April 25-27; Houston,
Texas, USA. Richardson, Texas: OnePetro. doi: 10.2118/148333-MS.
15. Sawaryn SJ, Grames KN, Whelehan OP. The analysis and prediction of electric submersible
pump failures in the Milne point field, Alaska. SPE Prod Fac. 2002; 17: 53-61.
16. Sawaryn SJ, Ziegel E. Statistical assessment and management of uncertainty in the number of
electric-submersible pump failures in a field. Proceedings of SPE Annual Technical Conference
and Exhibition; 2001 September 30-October 3; New Orleans, Louisiana, USA. Richardson, Texas:
OnePetro. doi: 10.2118/71551-MS.
17. Sawaryn SJ. The dynamics of electrical-submersible-pump populations and the implication for
dual-ESP systems. SPE Prod Fac. 2003; 18: 236-246.
18. Liu Y, Yao K, Liu S, Raghavendra CS, Lenz TL, Olabinjo L, et al. Failure prediction for rod pump
artificial lift systems. Proceedings of SPE Western Regional Meeting; 2010 May 27-29; Anaheim,
California, USA. Richardson, Texas: OnePetro. doi: 10.2118/133545-MS.
19. Takacs G. Electrical submersible pumps manual: Design, operations, and maintenance. Gulf
professional publishing; 2017.
20. Williams AJ, Cudmore J, Beattie S. ESP monitoring – Where's your speedometer? Proceedings
of 7th European Electric Submersible Pump Round Table, Aberdeen Section Table; 2003 April
30-May 2; Houston, Texas, USA.
21. Sherif S, Adenike O, Obehi E, Funso A, Eyituoyo B. Predictive data analytics for effective electric
submersible pump management. Proceedings of SPE Nigeria Annual International Conference
and Exhibition; 2019 August 5-7; Lagos, Nigeria. Richardson, Texas: OnePetro. doi:
10.2118/198759-MS.
22. Gupta S, Nikolaou M, Saputelli L, Bravo C. ESP health monitoring KPI: A real-time predictive
analytics application. Proceedings of SPE Intelligent Energy International Conference and
Exhibition; 2016 September 6-8; Aberdeen, Scotland, UK. Richardson, Texas: OnePetro. doi:
10.2118/181009-MS.

Page 18/19
JEPT 2022; 4(4), doi:10.21926/jept.2204036

23. Adesanwo M, Denney T, Lazarus S, Bello O. Prescriptive-based decision support system for
online real-time electrical submersible pump operations management. Proceedings of SPE
Intelligent Energy International Conference and Exhibition; 2016 September 6-8; Aberdeen,
Scotland, UK. Richardson, Texas: OnePetro. doi: 10.2118/181013-MS.
24. Bermudez F, Carvajal GA, Moricca G, Dhar J, Md Adam F, Al-Jasmi A, et al. Fuzzy logic application
to monitor and predict unexpected behavior in electric submersible pumps (part of the KwiDF
project). Proceedings of SPE Intelligent Energy Conference & Exhibition; 2014 April 1-3; Utrecht,
The Netherlands. Richardson, Texas: OnePetro. doi: 10.2118/167820-MS.
25. Grassian D, Bahatem M, Scott T, Olsen D. Application of a fuzzy expert system to analyze and
anticipate ESP failure modes. Proceedings of Abu Dhabi International Petroleum Exhibition &
Conference; 2017 November 13-16; Abu Dhabi, UAE. Richardson, Texas: OnePetro. doi:
10.2118/188305-MS.
26. Pennel M, Hsiung J, Putcha VB. Detecting failures and optimizing performance in artificial lift
using machine learning models. Proceedings of SPE Western Regional Meeting; 2018 April 22-
26. Garden Grove, California, USA. Richardson, Texas: OnePetro. doi: 10.2118/190090-MS.
27. Mubarak HA, Khan FA, Oskay MM. ESP failures / analysis / solutions in divided zone – case study.
Proceedings of Middle East Oil Show; 2003 June 9-12; Bahrain. Richardson, Texas: OnePetro.
doi: 10.2118/81488-MS.
28. Al-Sadah H. ESP data analysis to enhance electrical submersible pump run life at Saudi Arabian
fields. Proceedings of SPE Middle East Artificial Lift Conference and Exhibition; 2014 November
26-27; Manama, Bahrain. Richardson, Texas: OnePetro. doi: 10.2118/173703-MS.
29. Betonico GC, Bannwart AC, Ganzarolli MM. Determination of the temperature distribution of
ESP motors under variable conditions of flow rate and loading. J Pet Sci Eng. 2015; 129: 110-
120.
30. Lastra R. The quest for the ultrareliable ESP. Proceedings of SPE Middle East Artificial Lift
Conference and Exhibition; 2016 November 30-December 1; Manama, Kingdom of Bahrain.
Richardson, Texas: OnePetro. doi: 10.2118/184169-MS.
31. Birolini A. Reliability engineering: Theory and practice. Springer Science & Business Media; 2013.

Enjoy JEPT by:

1. Submitting a manuscript
2. Joining in volunteer reviewer bank
3. Joining Editorial Board
4. Guest editing a special issue

JEPT For more details, please visit:

http://www.lidsen.com/journal/jept

Page 19/19