Use of ESP Equipment in Special Conditions

CHAPTER 4
Use of ESP Equipment in

Special Conditions
Contents
4.1 Introduction 119
4.2 Pumping Viscous Liquids 120
4.2.1 Introduction 120
4.2.2 The Hydraulic Institute Model 121
4.2.3 Other Models 125
4.3 Production of Gassy Wells 128
4.3.2 Free Gas Volume Calculations 130
4.3.3 Pump Performance Degradation 136
4.3.4 Possible Solutions 140
4.3.5 Conclusions 154
4.4 Production of Abrasive Solids 155
4.4.2 Characteristics of Abrasive Materials 156
4.4.3 Sand Problem Areas 158
4.4.4 Solutions 160
4.4.5 Conclusions 164
4.5 High Well Temperatures 166
4.6 Variable Frequency Operation 167
4.6.2 Variable Speed Drives 170
4.6.3 Variable Frequency Generators 177
4.6.4 Interaction of VSD/VFG and ESP Units 178
4.6.5 Benefits of Using VSD/VFG Units 183
References 184
4.1 INTRODUCTION
The conventional ESP installation described in detail in the previous
chapter cannot be applied to many types of oil wells if operating condi-
tions differ from those ideal for the operation of the centrifugal pump.
Electrical Submersible Pumps Manual ISBN 978-1-85617-557-9

# 2009 Elsevier Inc. All rights reserved.
119
120 Gabor Takacs
This is because the ESP pump, due to its operational principle, can be best
used to lift single-phase liquids of low viscosity at relatively large flow rates.
Since the operation of the whole ESP system is essentially determined by the
performance of its pump, less-than-optimal conditions for the pump may
very adversely affect the effectiveness of this artificial lifting technology.
During the long history of the application of ESP systems, different solu-
tions to overcome the adverse conditions for centrifugal pumps have been
developed and this chapter is devoted to the description of the methods
available today. These solutions may involve the introduction of novel
equipment or procedures, or both, all with the final objective of stretching
the application ranges of ESP systems. The topics covered in this chapter
include the description of the performance of ESP pumps when pumping
highly viscous liquids, fluids with free gas, or abrasive solid contents. A spe-
cial section is devoted to installations using variable speed drive (VSD) units
which enable the operator to eliminate the ESP system’s greatest inherent
weakness: the ESP pump’s narrow application range in liquid rates.
4.2 PUMPING VISCOUS LIQUIDS

4.2.1 Introduction
As detailed earlier, ESP pump performance curves are obtained experi-
mentally by using water as a test fluid. Therefore, the design and analysis
of ESP installations based on these curves can be properly accomplished
only if water or a light crude is produced by the well. However, often
ESP units are used to produce crude oils or emulsions with considerably
high viscosities. Liquids with viscosities much greater than that of the
water cause increased frictional losses and disk friction inside the centrifu-
gal pump’s stages, resulting in lower developed heads, correspondingly
decreased pump efficiencies and high brake horsepower requirements. It
should be mentioned that performance parameters at shut-off conditions
(zero liquid rate) are independent of liquid viscosity. The main effects of
increased liquid viscosity are the following:
• pump capacity rapidly drops,
• the head developed by the pump stage also decreases,
• the required power to drive the pump increases, and
• the pump’s efficiency is reduced.
It follows from this that standard performance curves obtained from tests
made with water as a transported medium cannot be used with the required
accuracy if viscous liquids are handled by the ESP installation. If under such
Use of ESP Equipment in Special Conditions 121
conditions viscosity is not taken into consideration when selecting the

equipment, the ESP unit would be extremely overloaded. Although the
complete effect of liquid viscosity on centrifugal pump performance is not
yet known, several empirical methods exist for converting standard pump
performance curves, if measured ones are not available. The following sec-
tions will describe the most common solutions of this problem.
The traditional procedures of the conversion of water performance
curves do not account for the effect of pump stage geometry and utilize
uniform conversion factors for all geometries [1]. Also, the increase of
flowing temperature due to the heat generated by the pump is usually dis-
regarded. Therefore, in cases when oils of greater viscosity have to be
produced, the conventional procedures of viscosity correction detailed in
the following should be treated with caution.
Special problems arise when crudes with high water-cuts are produced
where the viscosity of the mixture may be much greater than that of the
crude oil alone. Such emulsions may behave like non-Newtonian fluids
and the description of their flow performance may be very difficult. In
such cases, complete performance testing of the ESP pump using the actual
liquid is advised.
4.2.2 The Hydraulic Institute Model
For the conversion of centrifugal pump performance curves to viscosities
greater than that of the water, Stepanoff [2] presented the first calculation
procedure but the most frequently used model was developed by the Hydrau-
lic Institute. The empirical procedure suggested in their publication [3] is based
on experimental data and provides a means to determine the centrifugal
pump’s performance for handling viscous liquids based on the pump’s per-
formance in water. The conversion involves the use of several diagrams and
is valid in the following ranges of parameters, although it is universally used:
• pump capacities (liquid rates) in the range of 3,400 bpd to
340,000 bpd,
• pumping heads between 6 ft and 600 ft, and
• liquid viscosities from 4 to 3,000 cSt (40–15,000 SSU).
The conversion of water performance parameters is based on the determi-
nation of correction factors and the following formulas:
Qvisc ¼ CQ Qw ð4:1Þ
Hvisc ¼ CH Hw ð4:2Þ
visc ¼ C w ð4:3Þ
122 Gabor Takacs
where: Qvisc, Hvisc, visc ¼ rate, head and efficiency for the viscous case
Qw, Hw, w ¼ rate, head and efficiency for the water case
CQ, CH, C ¼ rate, head and efficiency correction factors.
The calculation of the correction factors involves visual reading of two
diagrams, which is a time-consuming and inaccurate procedure. In order
to improve pump sizing for viscous liquids, Turzo et al. [4] curve-fitted
the original Hydraulic Institute diagrams and presented the numerical
model detailed in the following.
The correction factors are the sole functions of the corrected liquid rate,
Q , that depends on the viscosity of the liquid, the head and the rate
belonging to the pump’s best efficiency point (BEP) on the water perfor-
mance curves. The authors developed the following formulas to calculate
Q :

39:5276 þ 26:5605 lnðvÞ y
Q ¼ exp ð4:4Þ
51:6565
y ¼ 7:5946 þ 6:6504 lnðHwBEP Þ þ 12:8429 lnðQwBEP Þ ð4:5Þ
where: QwBEP ¼ water rate at BEP, 100 gpm

HwBEP ¼ head at BEP, ft
n ¼ liquid kinematic viscosity, cSt.
Based on the value of Q , the correction factors figuring in Eqs. 4.1–4.3
are determined from the equations given below. The Hydraulic Institute
model uses two single correction factors for the determination of the liq-
uid rate and the efficiency of the centrifugal pump in viscous service:
CQ ¼ 1:0 4:0327 103 Q 1:724 104 ðQ Þ2 ð4:6Þ
C ¼ 1:0 3:3075 102 Q þ 2:8875 104 ðQ Þ2 ð4:7Þ
The handling of the pump head is different and starts with selecting four
points on the original water performance curve at 60%, 80%, 100%, and
120% of the water rate belonging to the BEP. These four points are cor-
rected individually by using their respective correction factors, by using
the formulas:
CH0:6 ¼ 1:0 3:68 103 Q 4:36 105 ðQ Þ2 ð4:8Þ

CH0:8 ¼ 1:0 4:4723 103 Q 4:18 105 ðQ Þ2 ð4:9Þ
CH1:0 ¼ 1:0 7:00763 103 Q 1:41 105 ðQ Þ2 ð4:10Þ
CH1:2 ¼ 1:0 9:01 103 Q þ 1:31 105 ðQ Þ2 ð4:11Þ
At this point, using the calculated correction factors, the pump’s per-
formance curves can be plotted. Since viscosity at shut-in conditions
does not affect either the head or the efficiency, the original shut-in
head and efficiency can be directly plotted on the new curves. This
way, the pump head is known in five points which makes it possible
to estimate the head performance in a quite broad range of liquid flow
rates.
The brake horsepower required to drive the pump at any flow rate can
easily be calculated if the head, the liquid specific gravity and the pump
efficiency are known. The following generally applicable formula should
be used:
Q H gl
BHP ¼ 7:368 106 ð4:12Þ

where: Q ¼ pump capacity, bpd

H ¼ head developed by the pump, ft
gl ¼ liquid specific gravity, –
¼ pump efficiency, %.

Example 4.1
Calculate the head and efficiency curves of the pump whose water performance
curves for one stage are represented by the data given in the table below, if a liquid
with gl ¼ 0.9 and a viscosity of n ¼ 88 cSt is pumped.

Point Q bpd H ft Eff. %

0.6 QwBEP 540 27.9 50.9
0.8 QwBEP 720 25.5 60.3
QwBEP 900 21.8 64.0
1.2 QwBEP 1,080 15.2 55.4
124 Gabor Takacs
Solution
First, the corrected liquid rate is calculated, based on the value of y
given in Eq. 4.5, where QwBEP must be converted to 100 gpm (gallon
per minute) units as follows:
QwBEP ¼ 900 42=1440 ¼ 26:25 gpm ¼ 0:2625 100 gpm

y ¼ 7:5946 þ 6:6504 lnð21:8Þ þ 12:8429 lnð0:2625Þ ¼ 4:276
Now, using Eq. 4.4, Q is found:
Q∗ ¼ exp½ð39:5276 þ 26:5605 ln 88 þ 4:276Þ=51:6565 ¼ exp 3:15 ¼ 23:34
The correction factors are found from the corrected rate, Q , calculated
above, by using Eqs. 4.6–4.11:
CQ ¼ 1 4:0327 103 23:34 1:724 104 23:342 ¼ 0:812
C ¼ 1 3:3075 102 23:34 þ 2:8875 104 23:342 ¼ 0:385
CH0:6 ¼ 1 3:68 103 23:34 4:36 105 23:342 ¼ 0:890
CH0:8 ¼ 1 4:4723 103 23:34 4:18 105 23:342 ¼ 0:873
CH1:0 ¼ 1 7:00763 103 23:34 þ 1:41 105 23:342 ¼ 0:844
CH1:2 ¼ 1 9:01 102 23:34 þ 1:31 104 23:342 ¼ 0:797
The pumping rates valid for the viscous case are found from Eq. 4.1:
Qvisc1 ¼ 0.812 540 ¼ 438.5 bpd,
Qvisc2 ¼ 0.812 720 ¼ 584.6 bpd,
Qvisc3 ¼ 0.812 900 ¼ 730.8 bpd, and
Qvisc4 ¼ 0.812 1,080 ¼ 876.9 bpd.
The heads belonging to these liquid rates are calculated from Eq. 4.2:
Hvisc1 ¼ CH0.6 27.9 ¼ 0.89 27.9 ¼ 24.9,
Hvisc2 ¼ CH0.8 25.5 ¼ 0.873 25.5 ¼ 22.3,
Hvisc3 ¼ CH1.0 21.8 ¼ 0.844 21.8 ¼ 18.4, and
Hvisc4 ¼ CH1.2 15.2 ¼ 0.797 15.2 ¼ 12.2.
Finally, the pump efficiencies at the above liquid rates, according to Eq. 4.3:
visc1 ¼ 0.3854 50.9 ¼ 19.6,
visc2 ¼ 0.3854 60.3 ¼ 23.2,
visc3 ¼ 0.3854 64.0 ¼ 24.7, and

visc4 ¼ 0.3854 55.4 ¼ 21.3.
The calculated and the original performance curves are presented in
Fig. 4.1. As seen, pump efficiency radically decreases and the head devel-
oped by the pump also drops if crude with a higher viscosity is pumped.
4.2.3 Other Models
Some ESP equipment manufacturers developed viscosity corrections different
from the Hydraulic Institute model and recommended the use of tabulated
correction factors [5, 6]. The final correction of the performance curves is per-
formed similarly to the model previously discussed and Eqs. 4.1–4.3 are used.
An additional correction factor, CBHP, is introduced for calculating the brake
horsepower required for viscous service, and is used in the following formula:
BHPvisc ¼ CBHP BHPw gl ð4:13Þ
where: BHPvisc ¼ the required brake horsepower for the viscous case, HP
BHPw ¼ the required brake horsepower for the water case, HP
CBHP ¼ brake horsepower correction factor, and
g ¼ specific gravity of the produced liquid, –.
35 70
ηwater
30 60
25 50
Pump Efficiency, %
Head Developed, ft
Hvisc Hwater
20 40
15 30
ηvisc
10 20
5 10
0 0
0 200 400 600 800 1,000 1,200 1,400
Pumping Rate, bpd
Fig. 4.1 Pump performance curves for viscous service, Example 4.1.
126 Gabor Takacs
Table 4.1 presents the latest version [7] of the correction table that,
according to one manufacturer, provides standard approximations for
most cases. The table contains correction factors for pump capacity,
head, efficiency and brake horsepower, all in the function of liquid
Table 4.1 Viscosity correction factors according to Centrilift [7].
Viscosity SSU Correction factors

Capacity – Head – Efficiency – BHP –
50 1.000 1.000 0.945 1.058

80 0.980 0.990 0.870 1.115
100 0.970 0.985 0.825 1.158
150 0.947 0.970 0.736 1.248
200 0.924 0.958 0.674 1.341
300 0.886 0.933 0.566 1.460
400 0.847 0.909 0.497 1.549
500 0.819 0.897 0.462 1.590
600 0.792 0.883 0.434 1.611
700 0.766 0.868 0.410 1.622
800 0.745 0.858 0.390 1.639
900 0.727 0.846 0.368 1.671
1,000 0.708 0.833 0.349 1.690
1,500 0.659 0.799 0.307 1.715
2,000 0.621 0.771 0.272 1.760
2,500 0.590 0.750 0.245 1.806
3,000 0.562 0.733 0.218 1.890
4,000 0.518 0.702 0.278 2.043
5,000 0.479 0.677 0.149 2.176
Provided courtesy of Centrilift.

viscosity measured in SSU (Saybolt Seconds Universal) units. To convert

viscosities from the usual unit of centistokes (cSt) to SSU, use the follow-
ing formula:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
SSU ¼ 2:273ðcSt þ cSt 2 þ 158:4Þ ð4:14Þ

Example 4.2
Calculate the pump performance curves for the case described in Example 4.1, by
using the corrections given in Table 4.1. The brake horsepower values read at the four
rates from the water performance curves are 0.22, 0.225, 0.22, and 0.21 HP.

Solution
First, calculate the viscosity in SSU units from Eq. 4.14:
v ¼ 2:273½88 þ ð882 þ 158:4Þ0:5 ¼ 402 SSU
The correction factors are taken from Table 4.1 at a liquid viscosity of
400 SSU:
CQ ¼ 0:847; CH ¼ 0:909; C ¼ 0:497; CBHP ¼ 1:549:
Using the above correction factors, any point on the original pump perfor-
mance curves can be converted to the conditions of producing the given
viscous liquid. For example, the best efficiency point is modified as given
below, using Eqs. 4.1–4.3 and Eq. 4.13:
Qvisc ¼ 0.847 900 ¼ 762 bpd,
Hvisc ¼ 0.909 21.8 ¼ 19.8 ft,
visc ¼ 0.497 64 ¼ 31.8%, and
BHPvisc ¼ 1.549 0.22 ¼ 0.34 HP.
Figure 4.2 presents the calculated performance curves in bold line, where
the results of the previous example (utilizing the Hydraulic Institute
model) are shown in dashed line. As seen, the calculated heads are very
similar and the efficiency values are also in good agreement. Brake horse-
power values are plotted in Fig. 4.3, where the original water perfor-
mance curve is shown along with results of corrections by the Hydraulic
Institute and the Centrilift models.
128 Gabor Takacs
35 35
ηCL
30 30
25 25
ηHI
Pump Efficiency, %
Head Developed, ft
20 20
HCL
HHI
15 15
10 10
5 5
0 0
0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200
Pumping Rate, bpd
Fig. 4.2 Comparison of calculated viscous performance curves for Example 4.2.
0.4
HI Correction
0.4
Centrilift Correction
Brake Horsepower, HP
0.3
0.3
Water Performance
0.2
0.2
0.1
200 300 400 500 600 700 800 900 1,000 1,100 1,200
Pumping Rate, bpd
Fig. 4.3 Comparison of brake horsepower curves for Example 4.2.
4.3 PRODUCTION OF GASSY WELLS

4.3.1 Introduction
The ESP pump is a dynamic device that imparts a high rotational velocity
on the fluid entering the impeller, the amount of kinetic energy passed on
being proportional to the density of the fluid pumped. Because of their

great density, liquid particles receive a great amount of kinetic energy that,
after conversion in the pump stage, increases the flowing pressure. On the
other hand, although being subjected to the same high rotational speed,
free gas cannot produce the same amount of pressure increase because
the kinetic energy imparted on it by the impeller is significantly lower
due to the much lower density of the gas phase. Because of these reasons
the performance of centrifugal pumps always deteriorates if, along with the
liquid, free gas also enters the pump suction.
In spite of natural separation in the annulus and/or the operation of a
gas separator, gas may eventually reach the pump. Free gas in the ESP
pump rapidly ruins the pump’s efficiency and increased gas volumes may
cause fluctuations of pump output causing surges in well production. Surg-
ing in the pump leads to cyclic changes in motor load and the current
drawn by the motor starts to fluctuate accordingly forcing the surface
motor controller to shut down the ESP unit. Frequent system shutdowns
and restarts eventually damage the motor and the whole installation’s run
life is severely reduced.
The presence of free gas at pump intake and discharge involves addi-
tional implications in the design of the ESP installation.
• The two-phase mixture flowing from the intake through the pump
stages undergoes a continuous change in pressure that modifies fluid
properties like density and volume. The performance of the
subsequent pump stages, therefore, will be different, if compared to
a case where a single-phase liquid is pumped through all stages.
• In addition to the modified pump performance, the performance of
the well tubing changes, too, because gas evolving in the tubing above
the ESP unit decreases the average flowing density. This effect can
considerably reduce the required pump discharge pressure and an
appropriate correction in total dynamic head (TDH) calculations is
needed. In more sophisticated solutions like computer programs, tub-
ing pressure is calculated from a vertical multiphase pressure drop
model.
In ideal conditions, wells producing gassy fluids would be produced at
pump intake pressures above the well fluid’s bubblepoint pressure so there
is no free gas present at the pump suction. This would require a sufficiently
great submergence of the pump below the dynamic liquid level causing a
high flowing bottomhole pressure severely limiting the well’s production
rate. This is the reason why, in the majority of cases, specific solutions
(non-standard installation types, gas separators, etc.) have to be considered
when producing gassy fluids with ESP units.
130 Gabor Takacs
4.3.2 Free Gas Volume Calculations

The presence of free gas at the pump suction depends on the pump intake
pressure (PIP), the thermodynamic properties of well fluids and the well
temperature at suction conditions. The case of direct gas production, usually
a result of gas coning around the well, should also be mentioned. Depending
on the produced crude’s bubblepoint pressure, two cases are possible:
1. for PIP values higher or equal to bubblepoint pressure, no free gas
enters the pump, and
2. at PIP values lower than the bubblepoint pressure, a progressively
greater portion of solution gas evolves from the crude oil as free gas.
Since the amount of free gas increases the volumetric fluid rate the cen-
trifugal pump has to handle, the free gas as well as the total fluid volume
at pump suction conditions must be calculated by using the procedures
discussed in the following. It must be noted, however, that the actual
amount of gas that gets into the pump is usually less than the free gas
present at the pump suction. Depending on the characteristics of the
actual installation, part of the free gas can leave the annulus due to natural
gas separation.
The in-situ free gas volumetric rate present at pump suction conditions
is found from the well’s production gas/oil ratio (GOR) and the amount
of solution gas, Rs, still in solution at the pump intake pressure, PIP, as
given here:
q0g ¼ qo ðGOR Rs ÞBg ð4:15Þ
where: qg0 ¼ in-situ gas volumetric rate, ft3/d

qo ¼ oil volumetric rate, STB/d,
GOR ¼ production gas/oil ratio, scf/STB
Rs ¼ solution GOR at pump intake pressure, scf/STB
Bg ¼ gas volume factor at pump intake pressure, ft3/scf.
Since in the majority of cases no packer is set in ESP installations and the
annulus is open at the wellhead the annular space acts as a natural gas sep-
arator. In the stagnant liquid column existing above the pump suction gas
bubbles can rise to the dynamic liquid level and the gas thus separated
reaches the surface and enters the flowline. The natural separation of free
gas from the produced liquid in the annulus is a highly discussed topic in
the professional literature. Although a sufficiently accurate description of
the separation process is not yet available the model proposed by Alhanati
[8] can be used to approximate the amount of gas separated in the annulus.
Alhanati proposed the use of the following formula to estimate the
efficiency of natural gas separation:
vb
n ¼ 100 ð4:16Þ
vb þ vsl
where: vsl ¼ liquid superficial velocity, ft/sec,

vb ¼ terminal bubble rise velocity, ft/sec.
The liquid superficial velocity is the in-situ velocity of the liquid phase
inside the annulus and is determined from:

ql Bo WOR
vsl ¼ 6:5 105 þ Bw ð4:17Þ
A 1 þ WOR 1 þ WOR
where: ql ¼ liquid volumetric rate, STB/d

WOR ¼ production water/oil ratio, –
Bo ¼ oil volume factor at pump suction pressure, bbl/STB
Bw ¼ water volume factor at pump suction pressure, bbl/STB
A ¼ 0.0055 (IDc2 ODt2) annular area, ft2
IDc ¼ casing inside diameter, in
ODt ¼ tubing outside diameter, in.
The terminal velocity of the gas bubbles, vb, is found from the following
formula:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi 4 s gðrl rg Þ
Vb ¼ 2 ð4:18Þ
r2l
where: s ¼ interfacial tension, lb/sec2

g ¼ acceleration of gravity, ft/sec2
rl ¼ liquid density, lb/ft3
rg ¼ gas density, lb/ft3.
The densities of the phases are also calculated at pump suction condition as
follows:
0:0764 gg
rg ¼ ð4:19Þ
Bg
132 Gabor Takacs
where: gg ¼ specific gravity of the gas, –

Bg ¼ gas volume factor, ft3/scf.

g 1 g WOR
rl ¼ 62:4 o þ w ð4:20Þ
Bo 1 þ WOR Bw 1 þ WOR
where: go, gw ¼ specific gravities of the oil and water, –

Bo, Bw ¼ volume factors of oil and water, bbl/STB.
Based on the calculated value of the efficiency of the natural separation
process, the actual volume of the gas ingested by the pump is easily found
using the total gas volumetric rate calculated in Eq. 4.15. After converting
the gas rate to bpd units, we get:
qg n
0
0
qing ¼ 1 ð4:21Þ
5:61 100
The in-situ liquid volumetric rate entering the pump is found from the
water and oil rates:
0
ql ¼ qo ðBo þ Bw WORÞ ð4:22Þ
where: qo ¼ oil volumetric rate, STB/d

WOR ¼ production water/oil ratio, –
Bo ¼ oil volume factor at pump suction pressure, bbl/STB
Bw ¼ water volume factor at pump suction pressure, bbl/STB.
The total fluid volume the ESP pump has to handle is calculated as the
sum of the liquid and gas rates:
0 0 0
qt ¼ q l þ q ing ð4:23Þ
For the calculation of many of the terms defined previously, several ther-
modynamic properties of the liquid and gas phase must be known at the
pump suction pressure. However, measured thermodynamic parameters
of the produced fluids are seldom available and the use of general oilfield
correlations is recommended. In the following, selected correlations are
described, but the reader is warned that many other methods are also avail-
able and for accurate results one should select those that give the least
amount of error for the given oil field.
The Standing bubblepoint pressure correlation [9] allows the calculation

of solution gas/oil ratios at the pump intake pressure:
1:205
PIP
R s ¼ gg ð4:24Þ
18 10y
where: y ¼ 0.00091 T 0.0125 API

T ¼ suction temperature, F
API ¼ API gravity of oil, –
gg ¼ specific gravity of the gas, –.
The gas volume factor is evaluated from the following formula:
Z Ta
Bg ¼ 0:0283 ð4:25Þ
PIP
where: PIP ¼ pump intake pressure, psia

Ta ¼ intake temperature, oR
Z ¼ gas deviation factor, –.
The deviation factor of natural gas mixtures can be calculated by several
methods; the simple Papay formula [10] follows:
3:52ppr 0:274p2pr
Z ¼1 0:9813Tpr þ ð4:26Þ
10 100:8157Tpr
The terms ppr ¼ PIP/ppc and Tpr ¼ Ta/Tpc are the pseudo-reduced pres-
sure and temperature, respectively, where the critical pressure and
temperature can be found from the Hankinson–Thomas–Phillips cor-
relation [11]:
Ppc ¼ 709:6 58:7 gg ð4:27Þ
Tpc ¼ 170:5 þ 307:3 gg ð4:28Þ
where: gg ¼ gas specific gravity, –.

The volume factor of oil can be obtained from the Standing [9] correlation:
Bo ¼ 0:972 þ 1:47 104 F 1:175 ð4:29Þ

134 Gabor Takacs
where: F ¼ Rs(gg/go)0.5 þ 1.25 T

go ¼ oil specific gravity, –
T ¼ suction temperature, oF.

Example 4.3
Calculate the gas, liquid and the total fluid volumes at the suction of an ESP pump
with the following well data by considering the natural separation of gas in the
annulus.

Oil API degree ¼ 30

PIP ¼ 1,000 psi
Oil Sp.Gr. ¼ 0.876
Oil rate ¼ 1,000 STB/d
Gas Sp.Gr. ¼ 0.6
Production GOR ¼ 300 scf/bbl
Suction temp. ¼ 150 F
WOR ¼ 2
Casing ID ¼ 6.331 in
Tubing OD ¼ 2.875 in
Solution
First, the solution gas/oil ratio at the pump suction is found from Eq. 4.24:
y ¼ 0:00091 150 0:0125 30 ¼ 0:238
Rs ¼ 0:6½1; 000=ð18 100:238 Þ1:205 ¼ 147 scf =bbl
Next the volume factor of the gas phase is calculated based on the thermo-
dynamic parameters of the gas. Pseudocritical parameters are found from
Eqs. 4.27 and 4.28:
Ppc ¼ 709:6 58:7 0:6 ¼ 674:4 psi

Tpc ¼ 170:5 þ 307:3 0:6 ¼ 354:9 R
The reduced state parameters are:
Ppr ¼ 1; 000=674:4 ¼ 1:483

and
Tpr ¼ ð150 þ 460Þ=354:9 ¼ 1:719
Deviation factor can now be calculated from Eq. 4.26:
Z ¼ 1 3:52 1:483=100:9813 1:719 þ 0:274 1:4832 =1008157 1:719 ¼ 0:871
Volume factor of gas is evaluated from Eq. 4.25:
Bg ¼ 0:0283 0:871ð150 þ 460Þ=1; 000 ¼ 0:015 ft3 =scf
The free gas volume can now be calculated from Eq. 4.15:
0
qg ¼ 1; 000ð300 147Þ0:015 ¼ 2; 265 scf =d
The volume factor of the oil phase is found from Eq. 4.29:
F ¼ 147ð0:6=0:876Þ0:5 þ 1:25 150 ¼ 309
Bo ¼ 0:972 þ 1:47 104 3091:175 ¼ 1:1
The next task is the determination of the natural separation efficiency.

First, find the superficial liquid velocity in the annulus. The annular
cross-sectional area is:
A ¼ 0:0055ð6:3312 2:8752 Þ ¼ 0:175 ft2
The superficial liquid velocity from Eq. 4.17, assuming Bw ¼ 1:
vsl ¼ 6:5E5 3; 000=0:175ð1:1=3 þ 2=3Þ ¼ 1:15 ft=sec
Gas and liquid densities from Eq. 4.19 and Eq. 4.20, respectively:
rg ¼ 0:0764 0:6=0:015 ¼ 3 lb=ft3
rl ¼ 62:4½0:876=ð1:1 3Þ þ 2=3 ¼ 58 lb=ft3

136 Gabor Takacs
The terminal bubble rise velocity can now be calculated from Eq. 4.18,
by assuming s ¼ 0.04 lb/sec2, and substituting g ¼ 32 ft/sec2:
1=4
vb ¼ 21=2 f½0:04ð58 3Þ32=582 g ¼ 0:54 ft=sec
The efficiency of natural gas separation in the annulus is found from Eq. 4.16:
n ¼ 0:54=ð0:54 þ 1:15Þ100 ¼ 32%
The gas volume ingested by the pump is calculated from Eq. 4.21:
qing0 ¼ 2; 265=5:61ð1 0:32Þ ¼ 274 bpd
Liquid volumetric rate entering the pump, by assuming Bw ¼ 1.0, is cal-

culated from Eq. 4.22:
0
ql ¼ 1; 000ð1:1 þ 1 2Þ3; 096 bpd
The total in-situ fluid volume to be handled by the ESP pumps is found
from Eq. 4.23 as:
0
qt ¼ 3; 096 þ 274 ¼ 3; 370 bpd
4.3.3 Pump Performance Degradation

4.3.3.1 Gas Interference in Centrifugal Pumps
Free gas entering the centrifugal pump’s suction affects the performance of
the pump in several ways. Based on experimental as well as theoretical
studies [12–15], the pump’s head performance curve typically changes as
shown in Fig. 4.4. Compared to single-phase liquid performance, the fol-
lowing observations can be made:
• There is a region of unstable pump operation where surging or head-
ing occurs because of the cyclic changes in mixture density due to the
irregular flow in the impeller. This usually happens to the left of the
pump’s best efficiency point (BEP).
• Depending on the gas content at suction conditions and on the suction
pressure, surging can lead to gas locking when the impeller is fully
blocked by gas and no pumping action takes place.
• Stable operating points (indicated by circles in Fig. 4.4) fall below the
head performance curve valid for single-phase liquid operation.
Fig. 4.4 General trend of head

performance degradation.
Liquid Performance
BEP
Head Developed
Surging
Multiphase Performance
Pumping Rate
All investigators agree that these problems originate from the performance
of the impeller and that the diffuser plays a negligible role when pumping
multiphase mixtures.
Studies of the impeller’s operation while pumping liquid with free gas
content showed that pump performance did not significantly change when
the free gas phase was evenly dispersed in the liquid. In this case the only
effect on pump performance is that the free gas volume entering the pump
increases the total fluid volume the pump must handle (see Example 4.3).
The operating point on the pump performance curve shifts to the right,
and the head developed by the pump consequently decreases. The head
curve of the ESP pump for gassy fluids, therefore, deteriorates even for
small amounts of free gas evenly dispersed in the liquid phase.
In actual situations, however, flow in the impeller is inhomogeneous,
caused by the great difference between the densities of the two phases.
Gravity and centrifugal forces can therefore segregate the gas and liquid
particles more quickly than turbulence can mix them. This is the reason
why a homogeneous flow cannot be maintained and gas and liquid parti-
cles have a tendency to move inside the impeller independent of each
other. Gas, being the lighter phase, accumulates in pockets on the low
pressure sides of the impeller vanes because it does not have the sufficient
pressure to move to points at higher pressure. If these pockets are not
transferred toward the impeller discharge at a sufficient rate, they will grow
in size and can finally block the liquid flow through the impeller until gas
lock occurs.
138 Gabor Takacs
As described previously, most of the problems with free gas come from
the segregation of the phases in the pump’s impeller. Phase segregation is
affected by a host of parameters:
• Stage geometry. The free gas handling ability of a centrifugal pump
depends on its specific speed, Ns. The higher the pump’s specific
speed, the higher the amount of free gas it can manage. Thus radial
discharge pumps with low specific speeds are much more likely to
have gas problems than those with mixed flow stages having much
higher specific speeds. Axial pumps, on the other hand, are the best
in handling free gas. Also, pump stage geometries with low NPSH
net positive suction head (NPSH) values can handle greater free gas
volumes without gas locking.
• Bubble size. The size of gas bubbles has a great impact on the drag
and the buoyant forces acting on the bubbles immersed in the liquid
phase. Drag forces try to keep gas bubbles moving with the liquid,
whereas buoyancy forces try to separate them from the liquid. Smaller
bubbles tend to flow with the liquid phase and finely dispersed bubbles
are much harder to separate.
• Phase densities. The individual densities of the liquid and gas phases,
especially their ratio, determine the magnitude of buoyancy forces that
cause the phases to separate. The closer the densities to each other the
lower the danger of phase separation.
• Liquid viscosity. Liquid viscosity impacts on the drag forces acting
on the gas bubbles and the greater forces occurring at higher liquid
viscosities work against phase segregation.
• Pump speed. The pump’s rotational speed has a dual effect because
its increase builds up centrifugal forces trying to segregate the phases
but at the same time it also increases turbulence which reduces the size
of the gas bubbles and disperses them in the liquid.
4.3.3.2 Performance Criteria
When investigating the ability of ESP pumps to handle gas, Dunbar [16] and
Lea et al. [17] agree that pump suction pressure (a.k.a. pump intake pressure,
PIP) plays a great role in determining the amount of gas that can be ac-
commodated without a significant degradation of pump performance. The
widely used Turpin correlation [17] relates pump performance to the in-situ
gas and liquid volumes and the pump intake pressure (PIP). According to
Turpin, the limits of stable pump operation can be evaluated based on the
value of the following group:
0
qg
2; 000 0
ql
F¼ ð4:30Þ
3 PIP
where: qg0 ¼ gas volumetric rate at suction conditions, bpd

ql0 ¼ liquid volumetric rate at suction conditions, bpd
PIP ¼ pump intake pressure, psia.
Stable pump operation can be expected for values F < 1.0, while severe
gas interference and deterioration of pump performance occurs for cases
when F > 1.0.
The Turpin correlation is depicted in Fig. 4.5 where instead of the gas/
liquid ratio figuring in Eq. 4.30, the gas void fraction (the percentage of
free gas in the total fluid) is plotted versus the PIP. The areas of stable
and unstable operation are separated by a curve representing Eq. 4.30
for F ¼ 1.0. As seen in the figure, the amount of free gas that can be han-
dled by an ESP pump increases with increasing suction pressures (PIP).

Example 4.4
Using the Turpin correlation, decide if the gas volume calculated in the previous
example can be handled by an ESP pump.

Solution
The in-situ gas void fraction in percentages is easily found as:
void ¼ 359=3; 096 ¼ 11:6%
75
70
65
60
Gas Void Fraction, %
55 Unstable Operation
50
45
40
35
30 Stable Operation
25
20
15
10
5
0
0 200 400 600 800 1,000 1,200 1,400 1,600
Pump Intake Pressure, psia
Fig. 4.5 The Turpin correlation.
140 Gabor Takacs
Plotting this value at the pump intake pressure of 1,000 psi in Fig. 4.5
shows that the point falls in the safe operating area, so no gas handling pro-
blems are expected.
4.3.4 Possible Solutions

The many available solutions for fighting gas interference in ESP opera-
tions will be described in this section. The three main solutions [18] and
the ideas behind them are:
• Do not to allow free gas to enter the ESP system by utilizing the
natural separation of the liquid and gas phases in the casing annulus.
• If natural separation is insufficient and free gas enters the ESP system,
separate and expel free gas into the annulus before it can get into the
pump. Rotary gas separators are used for this purpose.
• If gas cannot be separated upstream of the pump, use special pumps
that can handle free gas along with the liquid without severe degrada-
tion of pumping action.
The necessary installation types and specific pieces of equipment will be
discussed in the next sections according to this classification [19, 20].
4.3.4.1 Utilization of Natural Gas Separation

In theory, the reverse flow gas separator discussed in Section 3.5 also
belongs to this category of gas avoidance in ESP systems. As detailed ear-
lier, this separator has a very narrow application range; this is why solutions
capable of removing considerable amounts of free gas from the wellstream
are discussed in the following.
4.3.4.1.1 Pump Set below the Perforations The use of a standard ESP
installation (with or without a gas separator), as depicted in Fig. 3.1, but
run below the well perforations can improve the performance of the ESP sys-
tem in gassy wells. By running the unit below the perforations the natural
separation of liquid and gas in the casing/tubing annulus can be fully utilized,
as in a natural gas anchor used in sucker rod pumping installations. Separation
by gravity occurs if fluid downward velocity is lower than the rising velocity
of the gas bubbles, usually at 0.5 ft/sec. This can be ensured only if casing and
equipment sizes allow, in which case free gas is automatically directed to
the surface without entering the pump suction. In addition to this effect,
the pump intake pressure (PIP) increases due to the greater pump submer-
gence causing the amount of free gas to decrease or even diminish. This
way the ESP pump’s efficiency will not be affected.
The basic prerequisite of this solution is the existence of a rathole or sump

in the well—that is, the perforations being situated above the total well depth
[21]. The serious drawback of placing the motor below the perforations is the
excessive heating of the motor because of the lack of liquid flow along the
ESP unit. To overcome this problem, one can use high temperature motors
that can be effectively cooled by the stagnant fluid and the formation. The
other remedy is shown in Fig. 4.6, where a smaller auxiliary pump, situated
between the protector and the ESP pump, redirects a part of the well’s liquid
production below the depth of the motor. Flow of the redirected liquid
causes movement of well fluids along the motor in the upward direction
and provides the cooling effect required for trouble-free operation.
4.3.4.1.2 Use of Motor Shrouds Motor shrouds (short sections of pipe

around the length of the ESP unit) have been successfully used to:
• act as simple reverse flow gas separators which, by changing the direc-
tion of flow, allow the buoyancy effect to decrease the amount of free
gas that enters the pump, and
• provide liquid flow along the ESP motor’s length to ensure proper
cooling of the unit.
Fig. 4.6 ESP installation using an auxiliary

pump to cool the motor.
Perforations
Pump
Aux. Pump
Protector
Motor
142 Gabor Takacs
Fig. 4.7 Typical shrouded ESP installation.
Perforations
Pump
Protector
Shroud
Motor
The simplest “open ended” shrouded installation is depicted in Fig. 4.7,

where the ESP unit is run below the perforations. The motor shroud (a
pipe section closed at the top and placed around the ESP unit) is hanging
from above the pump intake and forces well fluids to flow downward in
the casing/shroud annulus. The annular space must have a sufficiently large
cross-sectional area to ensure a low (preferably less than 0.5 ft/sec) fluid
velocity so that gas bubbles may rise and be vented up the casing annulus.
The shroud also guarantees that produced fluids flow along the motor’s
length thus providing proper cooling.
In case of greater gas production rates, in addition to the motor shroud,
the use of a simple reverse flow gas separator (see Section 3.5) is advised
and the required installation is shown in Fig. 4.8. Here the shroud is
installed just above the separator intake holes and vent tubes direct the
separated gas in the casing/tubing annulus above the perforations.
Often a dip tube is connected to the bottom of the regular motor
shroud, as seen in Fig. 4.9. The benefits of this design, as compared to
the use of open ended shrouds are:
Fig. 4.8 Shrouded ESP installation with gas

separator.
Perforations
Pump
Vent Tubes
Gas Separator
Protector
Shroud
Motor
Fig. 4.9 Shrouded ESP installation with

a dip tube.
Pump
Protector
Shroud
Motor
Dip Tube
Perforations
144 Gabor Takacs
• the natural separation of the free gas and liquid is greatly improved
because of the increased annular cross-sectional area available for
downward flow between the casing and the dip tube,
• well fluids can be produced from a restricted section of a vertical or
inclined hole where the ESP unit would not pass, and
• this arrangement can also be used in horizontal wells with the ESP
unit run in the vertical section and the dip tube reaching into the hor-
izontal part of the well.
An inverted shroud means a motor shroud open at the top. The shroud is
fixed below the pump intake and acts as a reverse flow gas separator, as
shown in Fig. 4.10. The ESP unit must be run above the perforations
so that the inverted shroud forces well fluids to flow downward in the
shroud/unit annular space. Proper design of the shroud diameter ensures
that flow velocity here is lower than 0.5 ft/sec required for the gravita-
tional separation of the gas from the liquid. The reverse flow velocity in
the annulus between the shroud and the ESP unit can be easily controlled
by installing at the top of the shroud a swage of a different diameter than
that of the shroud. Use of this type of ESP installation is advantageous in
horizontal wells with severe slugging problems because a long inverted
Fig. 4.10 ESP installation with an

inverted shroud.
Pump
Shroud
Protector
Motor
Perforations
shroud acts as a fluid reservoir that keeps the pump primed in periods
when large gas slugs are produced by the well.
4.3.4.2 Rotary Gas Separators
Rotary gas separators (RGSs) work on the principle that a multiphase mix-
ture, if spun at a high speed in a vessel, is separated to its constituent liquid
and gas phases due to the different levels of centrifugal force acting on the
liquid and gas particles. The rotational speed is provided by the separator’s
shaft, connected to the motor, and separation takes place in the body of
the separator. Here liquid is forced to the inner wall of the separator while
gas is concentrated near the shaft. A flow divider ensures that the separated
phases move along different paths and a crossover device directs (a) gas into
the casing annulus for venting to the surface and (b) liquid to the pump
intake.
4.3.4.2.1 Available Types The first rotary gas separator called the “pad-
dle-wheel” type appeared in the early 1970s [22] and is shown in
Fig. 4.11. It contains (usually five) axial vanes that run parallel along the
length of the separator’s shaft. Well fluid containing free gas is sucked in
Fig. 4.11 Rotary gas separator of the

Liquid paddle-wheel type.
Discharge
Gas
Discharge
Flow
Divider
Paddle
Wheel
Impeller
Well Fluid
Screen Intake
146 Gabor Takacs
the separator body at the bottom and enters the chamber containing the
rotating paddle-wheel impellers. The high centrifugal forces acting on
the liquid particles force them to move toward the separator’s wall, while
gas collects near the shaft due to the much lower forces acting on it. The
separated liquid and gas streams are split by the flow divider and after pass-
ing the crossover device, gas leaves to the annulus, while liquid reaches the
suction of the ESP pump.
Although the paddle-wheel separator provided a superior performance
in comparison to the reverse flow gas separator, discussed in Section
3.5, and is still in use today, it has a severe operational weakness that limits
its efficiency. This comes from the fact that the tips of the impellers, turn-
ing at a high speed, pick up part of the liquid from the inner wall of the
separator body and mix it with the less dense fluid situated closer to the
shaft. This remixing is inevitable since liquid velocity at the separator wall
is close to zero. This is the reason why separation efficiencies decrease in
wells with higher free gas contents. Another disadvantage of this design
is the abrasion between the tips of the impellers and the separator wall,
especially when pumping well fluids containing sand.
The rotating chamber type of gas separator [23] eliminates the remixing
of liquid and gas phases by isolating the rotating impellers from the stag-
nant liquid layer present on the inside wall of the separator body. As
shown in Fig. 4.12, the four impellers are enclosed by a rotor shroud.
Thus four separation chambers are created where the fluid rotates as a solid
Fig. 4.12 Different rotary gas

separator systems.
Paddle Rotating
Wheel Chamber Vortex
body and shearing and turbulence effects responsible for remixing of the
phases are minimized.
The multiphase mixture enters the separation chambers where centrifu-
gal accelerations of up to six times the acceleration of gravity ensure the
separation of the phases. A perfect separation would be possible if a suffi-
ciently long retention time could be maintained—that is, if the fluid stayed
in the separator chambers for a long enough time. By maximizing the
cross-sectional area of the chambers, axial fluid velocity can be held at a
low level and the retention time is increased accordingly.
At higher fluid rates, of course, fluid residence time in the separator
decreases and this inevitably involves a drop in separation efficiency. The
rotating chamber gas separator is good for wells producing high liquid rates
and/or highly viscous liquids because it involves the maximum accelera-
tion possible for an effective gas/liquid separation.
The vortex separator, shown in Fig. 4.12, is a very simple device that
has a single axial flow impeller as the only active member. The fluid enter-
ing the separator is spun by this impeller inducing a vortex in the other-
wise empty separator chamber. The vortex thus formed forces liquid to
move to the separator wall but gas stays near the shaft. The separated
phases are then divided and led to the annulus and the pump, just like in
the other separator types. Centrifugal forces in the separator chamber are
lower than in other rotary separators because the vortex generated in
the separator body spins at lower speeds than that of the separator shaft.
The efficiency lost this way is regained by the complete elimination of the
remixing of the gas and liquid phases present in other devices. The vortex
separator can be successfully used in wells producing sand but is not so
effective in viscous fluids and emulsions.
Versions of the rotary gas separators discussed so far include the one dis-
played in Fig. 4.13 [24]. Although the separation takes place due to the
operation of the impellers, the inducer has the very important function
of sufficiently raising the fluid pressure to move well fluids through the
separator. As found from extensive laboratory and field experiments, the
use of this rotary separator ensures ideal pump performance for in-situ
gas/liquid ratios as high as 0.6 [25].
Modern-day rotary gas separators typically include the following three
principal components:
1. the inducer (an axial centrifugal pump with a low NPSH value) that
increases the pressure of the incoming multiphase mixture,
148 Gabor Takacs
Fig. 4.13 Rotary gas separator with Liquid

inducer and paddle wheels. Gas Discharge
Discharge
Flow
Divider
Impeller
Inducer
Well Fluid
Intake
2. the guide vanes that modify the route of the fluid coming from the
inducer into an axial direction, in order to reduce shock losses, and
3. a separator chamber where the actual phase separation takes place
with the help of any of the devices described in Fig. 4.12.
4.3.4.2.2 Separation Efficiencies In principle, the separation efficiency

of any rotary gas separator depends on:
• the retention time (the time the fluid stays in the separation chamber),
and
• the magnitude of turbulence that occurs in the separator and causes
remixing of the phases.
The basic requirements for an efficient rotary gas separator are, therefore,
the following:
• the inducer should develop the proper amount of head to transfer the
fluids from the separator into the annulus,
• the retention time should correspond to the properties of the well
fluids so that phase separation becomes as complete as possible, and
• there should be a minimum of turbulence caused in the separation
chamber to reduce remixing of the separated phases.
100
90
80
Centrilift 400 Series
Gas Void Fraction, %
70
RGS Performance
60
50 Efficiency = 90%
40 85%
80%
30
75%
20
10 70%
0
0 500 1,000 1,500 2,000 2,500 3,000
Liquid Rate at Intake, bpd
Fig. 4.14 Separation efficiency of Centrilift 400 series gas separators [19]. Provided
courtesy of Centrilift.
Today’s rotary gas separators (RGSs) are very effective in separating free gas
from the wellstream at suction conditions. The typical performance of a
major manufacturer’s product is illustrated in Fig. 4.14 [19], based on
laboratory measurements. As shown, increasing fluid rates decrease the
separation efficiency because of the great mixture velocities entering
the gas separator.
Early investigations on the efficiency of RGSs were misleading because
they treated the separator’s operation independently of the natural gravita-
tional separation taking place in the annulus. Alhanati et al. [26] were the
first to recognize that the separation processes in the annulus and in
the RGS are closely linked. The effectiveness of natural separation in the
annulus affects the amount of free gas entering the separator, while the rest
of the gas is leaving to the casinghead. The RGS, because it cannot
achieve a perfect separation of the two phases, always transfers part of
the incoming gas to the pump and part of the liquid into the annulus.
The liquid expelled through the separator’s gas discharge ports is recircu-
lated to the intake and during its downward travel can take some of the
free gas bubbles with it. The natural separation process in the annulus is
thus altered and this also changes the composition of the mixture entering
the separator. Therefore, the separation processes in the annulus and the
150 Gabor Takacs
RGS have to be studied simultaneously, if meaningful performance data

are desired.
The key for the proper description of the separation process lies in the
simulation of the hydraulic processes inside the RGS. In order to expel
fluids through the discharge ports, the pressure of the gassy mixture enter-
ing the separator has to be increased by the inducer. If the head generated
by the inducer is sufficient to overcome the pressure losses across the dis-
charge ports, the separator expels fluids into the annulus; otherwise no sep-
aration takes place in the rotary gas separator.
Laboratory investigations, later confirmed by field data, showed that
RGSs operate in two performance regions depending on liquid flow rate.
Starting at low rates, high separation efficiencies are attained with increas-
ing liquid rates, but above a limiting rate efficiency suddenly drops to zero.
As shown for an example case [26] in Fig. 4.15, the composite separation
efficiency of the RGS and the annulus is high at lower liquid rates, then
suddenly drops to the level that characterizes the natural separation in
the annulus. The reason for the sudden change lies in the overloading of
the inducer that cannot generate any head above the critical flow rate. This
is due to the increased flow velocity inside the separator making the oper-
ation of the RGS impossible.
As shown previously, the efficient operation of the RGS heavily depends
on the proper functioning of its inducer. Inducers are axial flow pumps with
helical blades and low NPSH values; their most important geometrical
properties being the diameter, the pitch length and pitch number. The per-
formance of the RGS can be improved by optimizing the geometry of its
inducer [27].
100
90
Separation Efficiency, %
80
70
60
50 Test Conditions
40 GLR = 100 scf/STB
30 PIP = 200 psi
20
10
0
0 300 600 900 1,200 1,500 1,800 2,100 2,400 2,700
Liquid Rate, bpd
Fig. 4.15 Separation efficiency of a rotary gas separator and the annulus [26].
4.3.4.3 Gas Handling

The previous sections described the way:
• free gas can be prevented from entering the ESP pump, and
• how free gas that has entered the system can be separated before get-
ting into the pump.
If none of these solutions can be applied or their application is not recom-
mended for some reason, free gas inevitably reaches the ESP pump. In such
cases the pump must, either by using some special piece of equipment or some
modifications, be able to handle the gas—that is, to move it along with the liq-
uid phase. This section describes the various available techniques that make it
possible to pump considerable amounts of free gas with an ESP pump.
4.3.4.3.1 Overstaged Pumps The earliest solution was to use more pump
stages than normally required, so as to compensate for the smaller heads
developed by the first few stages due to gas interference. The so-called
“overstaged” pumps eliminate the overloading of the upper pump stages
usually associated with free gas production by supplementing the less-than-
sufficient total dynamic head developed by the original pump. The utiliza-
tion of oversized pumps with flow capacities greater than the required liquid
rate can also help handle greater amounts of free gas through the pump. Both
solutions use identical pump stage types for making up the entire pump.
Their common drawback is that different stages in the pump usually operate
at different liquid rates which may be outside the recommended range of the
given pump, inevitably leading to mechanical failures.
4.3.4.3.2 Tapered Pumps A successful and energy efficient solution for

handling free gas is the use of tapered pumps. These are ESP pumps made
up of several (at least two) different stage designs with the capacities of the suc-
cessive stages decreasing upward. The idea behind this solution is to allow the
lower stages to compress gassy fluids and to continuously decrease the total
volumetric flow rate through the pump. These stages do not significantly
increase the pressure, only to the level so that free gas is compressed with part
of it going into solution with the oil. The density of the fluid discharged from
the high capacity stages increases and the volumetric rate decreases accord-
ingly. Therefore, upper stages receive a much reduced fluid rate and can be
used to develop the head necessary to lift the fluids to the surface.
The effectiveness of tapered pumps is the function of total fluid rate. At
low to medium rates, the gas handling capacity of the ESP pump is greatly
increased because higher capacity stages with high specific speeds can be
152 Gabor Takacs
used at the bottom of the pump. But if high rates are desired, the upper
stages inevitably have higher specific speeds and much lower improve-
ments can be achieved by “tapering” the pump.
The design of tapered pumps should ensure that all stages in the pump
operate inside their optimum capacity ranges and this requires the use of
computer programs [28, 29]. Such programs calculate for every stage the
density and volume of the fluid as well as the pressure increase developed
by the stage and other parameters. The accuracy of the tapered pump
design heavily relies on proper well data and if well conditions differ con-
siderably from the assumptions used for the design, some or all of the stages
may happen to operate outside their operating range, destroying all advan-
tages and damaging the whole installation.
4.3.4.3.3 Stage Recirculation Since radial discharge (pancake) pumps

are most likely to get gas locked, several modifications to stage design were
attempted to improve gas tolerance. All designs attempt the recirculation
of liquid in the pump stage [30] and try to:
• break up the gas pockets usually created in regular stages, and
• increase the homogenization of the fluid flowing through the stage.
Figure 4.16 shows one version where matching holes are drilled in the top
shroud of the impeller and in the bottom shroud of the diffuser. The holes
provide a path for the fluid to flow from the diffuser back into the impeller.
The flow thus created greatly reduces gas segregation caused by centrifugal
forces in the impeller. The circulating liquid flow can also eliminate the
gas pockets that continuously grow in the eye of regular impellers and even-
tually lead to a gas locked condition. The big disadvantage of such solutions is
the inherent reduction of pumping efficiency by 20–30% or more. The
lower efficiencies are caused by the reduction of the head developed by
the stage due to the partial recirculation of the impeller’s discharge to the
intake.
Fig. 4.16 Pump stage with impeller circulation.

4.3.4.3.4 Gas Handlers Special devices connected upstream of the

pump can be used to improve the ESP pump’s tolerance to free gas pro-
duction. A gas charger, a short lower-tandem pump with high-capacity
stages, can be added below the main pump. It pumps the gassy fluids
entering the pump suction and while doing so compresses the mixture
so that the fluid is lifted easier by the main pump. Gas blenders disperse
free gas in the liquid phase; the small gas bubbles thus created are carried
with the liquid instead of resulting in a gas lock situation.
A recent development from a leading manufacturer [31, 32] makes use
of a gas handler utilizing special pump stages originally devised for transfer-
ring multiphase mixtures. The Poseidon centrifugal pump stages shown in
Fig. 4.17 contain impellers with helico-axial vanes and diffusers providing
a smooth axial flow. These special stages ensure an almost homogeneous
distribution of gas particles in the fluid because:
• radial flow velocities responsible for gas segregation in impellers are
practically eliminated due to the low centrifugal force developed by
the axial flow impeller, and
• the stage effectively mixes the two phases.
The Poseidon gas handler is connected between the pump intake and the
ESP pump and acts as a charge or booster pump. The homogeneous mix-
ture of liquid and gas leaving the gas handler provides ideal conditions for
the ESP pump to lift the fluid to the surface. The unit can handle well-
streams with up to 75% of free gas content at the pump intake; effectively
Fig. 4.17 Schematics of Poseidon stages.
Diffuser
Impeller
154 Gabor Takacs
preventing gas locking of ESP pumps. Available models [32] work in the
flow rate range between 5,000 bpd and 9,000 bpd and need a substantial
power of 50 HP to operate.
4.3.4.3.5 Special Pumps Special ESP pumps designed to produce well

fluids with very high free gas content are also available. These are tapered
pumps with an intake inducer (axial flow pump) at the bottom, specially
developed, low NPSH (net positive suction head) pump stages in the mid-
dle, and standard stages at the top of the pump assembly. The intake inducer
provides a positive pressure for the gas/liquid mixture to enter the upper
stages, the low NPSH stages mix and compress the gassy fluid so that the
standard stages at the top can increase the flowing pressure to the level
required for lifting the liquid to the surface. Thanks to the special design
of the low NPSH stages, surging and gas locking are eliminated, thus ensur-
ing a long operational life of the ESP equipment even in gassy wells.
4.3.5 Conclusions
As seen from the preceding discussions, the production of gassy wells with
high free gas content is no longer outside the application range of the ESP
system. By using the proper equipment, ESP installations can be used to
lift oil wells with very high gas production rates. The possible solutions
include the proper choice of pump stage types, the use of tapered pumps,
RGSs and combinations of these.
Figure 4.18, published by a major manufacturer, compares the applica-
tion ranges of the different possible solutions of handling free gas in the
function of the gas void fraction valid at pump suction conditions [34].
Radial Flow Pump

Mixed Flow Pump
Tapered Pump
GasMaster Pump
Pump w. Separator
Tapered Pump w. Separator
Pump w. Tandem Separator
0 10 20 30 40 50 60 70 80 90 100
Gas Void Fraction at Pump Suction, %
Fig. 4.18 Comparison of free gas handling abilities of ESP systems [34]. Provided
courtesy of Centrilift.
Radial flow ESP pumps can only be used for low void fractions, but
pumps with mixed flow stages perform about equally with tapered pumps.
Special pumps (GasMaster) can handle up to 50% free gas in the total fluid
stream. The use of rotary gas separators (RGSs) very significantly increases
the operational range of the ESP installation. The greatest relative amounts
of free gas can be handled by a tandem RGS, composed of different types
of separators: reverse flow, or a version of the rotary gas separator.
4.4 PRODUCTION OF ABRASIVE SOLIDS

4.4.1 Introduction
The ESP pump is a high speed rotating device whose stationary and rotat-
ing parts are lubricated by the fluid pumped. Since oil well fluids very
often contain solid particles, the abrasive action of those can easily damage
the moving parts of the pump. Metal loss caused by abrasion and/or ero-
sion at critical points in the pump stage or in its different bearings (radial
and axial), even if other components of the pump are not attacked, may
lead to the catastrophic failure of the pump.
In the following, sand problems and their solutions will be shown for the
case of the ESP pump, since the great majority of sand problems in the ESP
system occur in the centrifugal pump. However, rotary gas separators are also
prone to sand damage and special separators have to be used in sandy wells.
In short, sand production severely shortens the life expectancy of the ESP
system; this is the reason why special technical solutions and the use of
sophisticated materials are necessary when pumping solids-laden fluids.
Although solids other than sand, such as iron sulfide, calcium carbonate,
and so on, may cause abrasion in centrifugal pumps, most of the abrasion
problems originate from the production of sand (quartz, SiO2) along with
well fluids. The characteristic features of sand production from oil wells
can be explained by the application of rock mechanics principles and can
be summed up as follows:
• sand production usually starts at high well rates,
• the amount of produced sand increases after water breakthrough to the
well occurs, and
• sand production increases when well flow rates change.
ESP units are usually installed in wells with high fluid rates and low bottom-
hole pressures, are often the prime choice in fields under waterflood
operations, and frequently involve cyclic operations. Since these conditions
exactly match the main causes of sand production in oil wells listed pre-
viously, ESP systems are particularly prone to sand problems.
156 Gabor Takacs
4.4.2 Characteristics of Abrasive Materials

The effect of sand-laden fluids flowing through the ESP pump is the
removal of metal particles from different parts of the pump. Damage can
be classified as:
• erosion occurring on a metal surface hit by abrasive material particles
contained in the fluid, and
• abrasion caused by an abrasive material and occurring between two
metal surfaces due to mechanical wear.
The magnitude of damage caused by these types of wear depends on many
factors: the hardness of the attacked metal and that of the solid particles,
the concentration, size, shape, toughness and particle size distribution of
the solid particles [35].
The hardness of the abrasive particles as compared to that of the
attacked metal has a direct effect on the damage made. Since no abrasive
will cut anything harder than itself, metal surfaces in ESP pumps used
for abrasive service must be harder than the abrasive contained in the well-
stream. Therefore, the proper choice of the base metal or the application
of a hard metal coating to endangered surfaces is recommended.
Figure 4.19 illustrates the relative hardness of several materials in ESP
service. Also shown is the Mohs scale used for comparing the hardness
of minerals; where a material with a higher Mohs rating will scratch any
Fig. 4.19 Approximate material hardness Diamond 10

vs the Mohs scale.
Silicon Carbide
Tungsten Carbide
9
8
Zirconia
Quartz (Sand) 7
Tool Steel
6
Nickel
5
Ni-Resist
Copper 4
Iron Sulfide
Calcium Carbonate 3
MOHS Scale
other material with a lower rating. As seen, sand or quartz (SiO2) is harder
than regular steel or nickel but cannot damage the much harder exotic
materials like tungsten carbide or zirconia. The materials of other solids
usually present in wellstreams (iron sulfate, calcium carbonate) are much
softer than sand; this is the reason why sand is considered the key abrasive
substance in oil wells.
Particle size and shape of the abrasive material are also important and
affect the damage done by abrasion or erosion. Abrasion is highest when
the size of solid particles is comparable to the clearances used in ESP
pumps; sizes between 50 and 250 microns are considered to be the most
damaging. On the other hand, wear due to erosion is proportional to par-
ticle size and the square of particle velocity. Rough, irregularly shaped
solid particles do more damage than rounded, smooth ones of the same
size.
Solids concentration in the produced fluid has a prime effect on the
damage done by erosion and abrasion in ESP equipment. When making
estimates of the abrasive production of a well, one must consider that sam-
ples taken from the wellstream may contain particles with relatively low
abrasive properties like salt crystals, corrosion products, scale, and so on.
To eliminate these, samples should be treated with concentrated acid to
dissolve the nondestructive materials and leave a representative sample
containing mostly sand. Sand concentration can be expressed in ppm (parts
per million) or its equivalent unit of milligrams per liter; the usual classifi-
cation of sand production is given in the following.
Concentration, mg/liter Definition

Less than 10 Light
11–50 Moderate
51–200 Heavy
More than 200 Severe
Considering the factors detailed previously, one can define an aggressive-

ness index [36] that shows the relative destructive power of the particular
sand sample and allows the comparison of different wellstreams. It is
expressed in percentages and represents the combined effects of particle
size, shape and sand content in the solids produced with higher aggressive-
ness indexes meaning more aggressive conditions. In lack of a detailed sand
analysis a value of 60% is usually assumed.
158 Gabor Takacs
4.4.3 Sand Problem Areas

Production of abrasive solids along with well fluids has the greatest effect
on the operation of the ESP pump where abrasive particles move with
high local velocities. Abrasive damage caused in ESP pumps takes different
forms and occurs in different parts of the pump [37]. Figure 4.20 illus-
trates, in the case of a fixed impeller pump, the possible wear points where
sand damage can be classified as:
• erosion in the pump stage,
• abrasion in radial bearings (radial wear), and
• abrasion in thrust washers and thrust bearings (axial wear).
In addition to pumps, other components of the ESP system are also
affected by abrasives in the well fluid. Rotary gas separators (RGSs) are
especially susceptible to abrasive wear because of the great centrifugal
forces acting on the solid particles [19]. The solids, hitting the separator
housing with great speed, remove the layer of corrosion products from
the metal surface that immediately corrodes again and the process repeats.
This eventually may lead to a complete cutting of the separator housing,
necessitating the use of special, abrasion resistant materials.
According to their great importance the different types of abrasive dam-
age occurring in the ESP pump will be detailed in the following.
4.4.3.1 Pump Erosion
Erosion in pump stages is caused by the abrasive solid particles striking the
metal surfaces just like in sandblasting. The wear caused is greater for large
and rough solid particles than for small and smooth ones. Also, wear is
proportional to the square of flow velocity because the destructive poten-
tial of the solids is related to their kinetic energy.
Fig. 4.20 Sand problem areas in a fixed

impeller pump.
Impeller Erosion
Abrasion
Diffuser Bearing
Abrasion
Because of the cushioning effect of the liquid phase the suspended solid
particles do most of the damage at points where a change of flow direction
takes place: at the entrances to the diffuser and to the impeller. Another
type of erosion occurs around the balance ring in the diffuser where solid
particles are moved by viscous drag in the stagnant fluid.
Although erosion of pump stages can be considerable, it seldom leads to
failure because the pump usually fails for other reasons long before it is
completely eroded.
4.4.3.2 Abrasion in Radial Bearings
Radial wear is caused by abrasion in the pump’s radial bearings. In standard
pumps, radial support of the pump shaft is provided by simple journal
bearings with the impeller hub acting as the journal and the diffuser bore
being the bearing. Journal and bearing materials are identical with that of
the stage with well fluid lubricating the two parts. These bearings run almost
for the total length of the shaft and have fixed radial clearances, the amount of
clearance depending on pump design and machining tolerances. Clearances
are usually small but large enough for the majority of sand particles to enter
the space between the bearing and the journal. Larger particles, after entering
the clearance space, are crushed and remove metal from the bearing surfaces,
while small ones may be taken by the fluid flow without even touching the
bearing surfaces. The amount of wear in radial bearings heavily increases
with increased flow rates because of the increased amount of sand particles
carried by the fluid.
The main effect of radial wear in an ESP pump is the growing of the
clearances in bearings and sleeves. This brings about the loss of radial sta-
bility of the shaft which starts to rotate eccentrically causing the side loads
in the bearings to increase and these, in turn, further accelerate wear.
Because of the pump shaft’s slenderness and the high axial loads acting
on it the shaft starts to buckle, inducing severe vibrations along the shaft.
These vibrations can completely destroy the pump in a very short time.
Since the pump shaft is directly connected to the protector, vibrations
are transmitted to the shaft seals protecting the electric motor from well
fluids. Eventual failure of these seals leads to a complete system
breakdown.
4.4.3.3 Abrasion in Thrust Washers
Axial wear is caused by abrasion in thrust bearings and occurs on the thrust
washers and the mating surfaces in the pump stage. Abrasive particles caught
between the washers result in worn washers or even in metal-to-metal
160 Gabor Takacs
rubbing of the impeller on the diffuser. Since pumps with fixed impellers
(a.k.a. compression pumps) completely eliminate the contact between
impellers and diffusers, they are almost totally protected against the effects
of abrasive fluid production.
In pumps with floating impellers the axial forces are absorbed by the thrust
washers of the impellers that are free to “float” depending on flow condi-
tions. The clearance, therefore, is not fixed but varies with the magnitude
of the thrust and the viscosity of the fluid. In the recommended capacity
range the pump is in the downthrust condition (see Fig. 3.4) and the clear-
ance thus created is usually too small for the majority of sand particles to enter
this space. However, if operating in the upthrust condition, the clearance
between the washers greatly increases allowing large grains of sand to enter,
and this can lead to completely worn-out washers and/or abraded stages.
4.4.4 Solutions
Over the years, manufacturers developed many modifications in pump
design and introduced the use of different materials for fighting sand dam-
age in ESP equipment. The common background for all designs is the use
(on all influenced points in the pump stage) of materials hard enough to
resist the harmful effects of abrasives. Since the most aggressive abrasive
material, almost always present in well fluids, is sand, all materials harder
than sand (see Fig. 4.19) can be used at critical points in the pump stage.
Interestingly enough, soft materials like rubber can also be successfully
used in journal bearings. In this case, due to the resilient nature of the
material, sand particles entering the clearance between the rubber bearing
and the metal journal while hitting the rubber do not remove any material
because of the rubber’s deflection. In addition to this, sand particles cannot
imbed in the soft rubber part. All these result in the sand particles working
their way out of the bearing thus greatly reducing abrasive wear on the
metal journal.
As discussed in the previous section, the severity of abrasive damage in
submersible pumps increases in the following order:
1. erosion in impellers and diffusers,
2. axial wear in thrust bearings and up- and downthrust washers in
floater pumps, and
3. radial wear in radial (journal) bearings.
Erosion wear in pump stages can be minimized by using special metals
(Ni-Resist, an alloy containing 18% nickel) for manufacturing of impellers
and diffusers, instead of the less expensive gray iron, or by using hard sur-
face coatings on endangered areas.
Axial abrasion is present in thrust bearings and the up- and downthrust
washers of floater pumps. In the ESP unit’s main thrust bearing, situated in
the protector, extremely hard materials like ceramics (usually zirconia) are
used for thrust runners and shoes. The wear of the washers used in floater
pumps can be reduced by increasing their surface areas and by the proper
selection of their materials.
Since radial abrasion is the most significant effect of sand damage in ESP
pumps, the various ways of reducing it are detailed in the following.
4.4.4.1 Reduction of Radial Wear
The earliest solution [37] to decrease radial wear was the placement of spe-
cial radial bearings at regular intervals in the submersible pump. Such bear-
ings contain a special resilient (usually rubber) bushing pressed into the
diffuser bore where the pump shaft turns, see Fig. 4.21. The rubber bear-
ing is fluted—that is, it has longitudinal grooves on its inside surface where
sand particles are washed into and are continuously removed from by the
fluid pumped. By fitting these bearings in several stages instead of the stan-
dard diffuser bore/impeller hub-type bearings, radial abrasion damage can
be reduced. The shorter the distance between the special bearings
(distributed evenly along the length of the pump shaft), the greater the
radial stability of the pump shaft becomes.
Hardening of wearing surfaces to decrease abrasion in radial and thrust
bearings was also applied. Figure 4.22 illustrates several stages of a floating
impeller pump modified for abrasive service [38]. Radial and axial
Diffuser Bore Rubber Bearing Fig. 4.21 Construction of a resilient

radial bearing.
Impeller
162 Gabor Takacs
Fig. 4.22 Floating impeller pump stages Axial Wear

modified for abrasive service.
Radial
Wear
stabilization is ensured by the use of special inserts made of materials of

great hardness so that radial and axial wear can be minimized.
A detailed view of the pump stage is given in Fig. 4.23, where a hard-
ened insert is fixed to the diffuser in which a flanged sleeve with a similar
hardness is turned by the pump shaft. The radial and axial wear surfaces
are, because of their great hardness, highly resistant to abrasion damage.
The great disadvantage of having all pump stages of the abrasion resis-
tant version is the extremely high cost; this is why floater pumps with only
a few abrasion resistant stages (usually situated at the top and at the bottom
of the pump) are also manufactured. Compression pumps with all or some
of the pump stages being of the abrasion resistant type are also available for
producing well fluids of extremely high solid contents.
Fig. 4.23 Modified pump stage used for

Impeller
abrasive service.
Flanged
Sleeve
Hardened
Insert
Diffuser
The use of special hard materials such as silicon carbide, tungsten car-
bide, or ceramics can greatly increase the abrasion resistance of ESP pump
parts. The application of these materials in ESP pump bearings, however,
proved to be unsuccessful because they are very brittle and are easily frac-
tured if loaded at one point or on a line. Regular journal bearings are
mounted into their housings by press fitting and this technique inevitably
causes line loadings and an eventual failure of the bearing if very brittle
materials are used. This is the reason why journal bearing designs had to
be improved to facilitate the utilization of extremely hard materials [39].
The compliant mounted journal bearing illustrated in Fig. 4.24 was
developed for use with extremely hard bearing materials (silicon carbide,
tungsten carbide, ceramics, etc.). The bearing is fitted into the housing
so that a fluid chamber is formed by the two O-rings. This chamber in
conjunction with the elastic O-rings acts as a vibration and shock damp-
ener and allows the bearing to find its best running position; point or line
loading of the bearing is thus avoided. Therefore, the compliant mounting
of journal bearings eliminates the inevitable failures due to the low fracture
strength of the extremely hard materials.
The usual material selections in compliant bearings are zirconia bearings
and journals, or zirconia bearings with silicon carbide journals. Zirconia is
a ceramic material of great hardness that is virtually unaffected by abrasives
in the wellstream and can withstand temperatures up to 1,000 F. It has
excellent lubrication properties and is not affected by the presence of
H2S or CO2 gases.
A submersible pump specifically designed for abrasive service is illus-
trated in Fig. 4.25 where head and base bearings (at the two ends of
Shaft
Retaining Fig. 4.24 The structure of a
Spacer Ring compliant radial bearing.
O-Ring
Compliant
Bearing
O-Ring
Journal
164 Gabor Takacs
Fig. 4.25 Constructional features of an

abrasion resistant pump.
Head
Bearing
Compliant
Bearings
Base
Bearing
the pump shaft) are of the compliant version. As shown in the figure, sev-
eral similar bearings are distributed along the pump shaft, their spacing,
due to their moderate cost, may be kept as close as possible. Closer spacing
results in lower shaft deflections and lower vibrations and a consequent
increase in radial stability. Radial instability is the prime cause of mechani-
cal failure in ESP pumps, and run life of ESP equipment in wells produc-
ing abrasives can greatly be increased by the use of compliant bearings.
4.4.5 Conclusions
Production of sand or other abrasive solid materials along with the well
fluid severely shortens the run life of regular ESP equipment, the most
affected component being the centrifugal pump. The ESP industry, how-
ever, offers different solutions to combat the harmful effects of sand in the
produced fluid. Thanks to special equipment components (mainly pumps),
the efficiency of ESP installations in sand producing wells can be main-

tained at levels competitive with other types of artificial lift.
As a guide to the selection of the proper ESP pump for abrasive service,
Fig. 4.26 is presented [36]. The application range of the different pump
types is shown in the function of the sand concentration and the aggres-
siveness index introduced previously. If a detailed sand analysis is not avail-
able, an aggressiveness index of 60% is assumed.
Depending on the severity of abrasiveness, the following general rules
apply to the selection of the proper ESP pump:
• Floating impeller pumps (floater pumps) are generally not recom-
mended if sand production is present.
• Under mildly abrasive conditions, compression-type pumps with fixed
impellers and without any special features can be used.
• Under aggressive well conditions, abrasion resistant (AR) (a.k.a. radi-
ally stabilized) floating pumps can be applied.
• Extremely aggressive conditions require the use of abrasion resistant
(AR) (a.k.a. radially stabilized) compression-type pumps.
The fight against abrasion is not finished by the selection of abrasion resis-
tant pumps because trouble-free operations require several basic rules to be
obeyed. System startup in abrasive wells should be specially conducted
to help the initial production of sand that has settled above the ESP unit
during the shutdown period. Floating impeller pumps should always be
operated in the downthrust region, and screens or filters, and so on must
be used.
100 Fig. 4.26 Pump selection

90 AR chart for abrasive service.
Compression
Aggressiveness Index, %
80 Pump
70
60 AR
50 Floater
Pump
Floater Pump
40 Compression
30 Pump
20
10
0
0 20 40 60 80 100 120 140
Sand Content, mg/liter
166 Gabor Takacs
4.5 HIGH WELL TEMPERATURES

Well temperature has always been a limiting factor in the application of
ESP units, standard ESP equipment can be applied to a maximum ambient
temperature of approximately 240 F. Above this limit, the performance of
average ESP components quickly deteriorates, and an eventual failure can
be expected. Since the depths of today’s oil wells are increasingly greater
where extreme well temperatures are common, the latest developments
aim to provide new materials and designs to ensure proper operation of
ESP installations at higher temperatures.
The main effects of high temperatures on the performance of submers-
ible equipment are diverse and can be classified as:
• elastomers used in ESP equipment for sealing, and so on are weakened
and their service life is shortened accordingly,
• the dielectric properties of insulation materials deteriorate and
motor or cable insulations may lose their dielectric strength causing
burnouts,
• the electrical resistance of the conductor in the submersible cable
increases causing an increased power loss in the cable,
• the viscosity of motor oil decreases and the load carrying capacity of
the main thrust bearing, usually situated in the protector section,
decreases,
• due to the different thermal expansion of dissimilar metals mechanical
failures in rotating machinery can occur, and
• scale formation on the inside and outside surfaces of ESP equipment
becomes more pronounced.
As regards operating temperature, the most critical component of the ESP
unit is the submersible motor because it not only operates under the ambi-
ent well temperature but the large amount of heat generated during oper-
ation raises its inside temperature. Overheating of ESP motors, in addition
to extreme well temperatures, may be caused by:
• motor overload,
• insufficient cooling caused by a low liquid flow velocity past the
motor; see Fig. 3.16,
• production of a fluid with unexpectedly low specific heat providing an
insufficient cooling effect,
• scale deposition on the outside of the motor severely reducing the
amount of heat dissipated to the well fluid, and
• the presence of harmonics in input power voltage, usually associated
with those VSDs (variable speed drives) providing quasi-sinusoidal
waveforms.
In order to limit the temperature rise and to increase the service life of stan-
dard ESP motors, the general solution is motor derating when a motor with a
higher than necessary rating is chosen. As discussed in Section 3.3.4.2, by
decreasing the horsepower load (the ratio of actual power to the rated power)
the heat generated in the motor can be decreased so that the operating tem-
perature falls below the motor’s rated temperature (see Fig. 3.18). Liquid
flow velocity past the motor should be checked also and if not sufficient, it
can be increased by using a larger diameter motor or a shroud attached to
the outside of the motor.
The main features of motors developed especially for high temperature
service are the following:
• windings are insulated by epoxy to increase the dissipation of heat
generated in the stator windings to the outside of the motor due to
the high heat conductivity of epoxy, as compared to simple varnish
coatings,
• rotating clearances are increased to provide for the larger thermal
expansion of the different metal parts occurring at higher tempera-
tures, and
• special outside coatings on ESP components reduce the severity of
deposition of scale or other precipitants and prevent the reduction of
the cooling effect of the wellstream.
In addition to the changes in motors, high temperature applications necessi-
tate several modifications in the protector, the pump and the cable. The
rotating members (protector, pump) should have increased inside clearances
just like the submersible motor. The seal material providing the best perfor-
mance at high temperatures is EPDM (ethylene propylene diene monomer)
and all elastomers in motors, pumps and protectors must be made of this
material. In high temperature electric cables the insulation and the jacket
are also made of EPDM materials.
The improvements discussed so far have highly increased the temperature
limits of ESP applications. Today all manufacturers offer ESP equipment for
operating temperatures of 400 F; reliable operation of ESP motors at 500 F
in a steam-flooded field was also reported [40].
4.6 VARIABLE FREQUENCY OPERATION

4.6.1 Introduction
In previous chapters it was often pointed out that the conventional ESP
system running at a constant speed is very inflexible, if compared to other
168 Gabor Takacs
types of artificial lift. The reason is that operating an ESP pump outside
its quite narrow recommended liquid rate range not only reduces the sys-
tem’s efficiency but can lead to early equipment failures. This is why the
proper knowledge of well inflow parameters is an absolute prerequisite of
a proper design and why the use of improperly assumed well per-
formance data results in bad designs. But even in cases with accurate data
and a perfect installation design, the inevitable changes in well inflow
parameters (formation pressure, fluid rates, etc.) with time can lead to a
quick deterioration of the operating conditions and an eventual system
breakdown.
The above problems, along with some other characteristic disadvantages
of conventional ESP installations, are eliminated if the submersible pump
were driven with widely variable speeds. Since the submersible motor
drives the centrifugal pump directly, pump speed is simply controlled by
changing the motor’s rotational speed. For a given motor construction,
however, motor speed is a direct function of the frequency of the AC cur-
rent, so a proper regulation of the power supply’s frequency achieves the
necessary effect. This is the background of the application of variable fre-
quency power operations in artificial lifting by ESP systems, the two ver-
sions of which are variable speed drives (VSDs) and variable frequency
generators (VFGs).
Since the submersible motor’s speed is directly proportional to the fre-
quency of the AC supply, the affinity laws describing the operation of the
ESP pump and introduced in Chapter 2 have to be expressed in the func-
tion of the frequency:

f2
Q2 ¼ Q1 ð4:31Þ
f1
2
f2
H2 ¼ H1 ð4:32Þ
f1
3
f2
BHP2 ¼ BHP1 ð4:33Þ
f1
where: f2, f1 ¼ AC frequencies, Hz

Q1, Q2 ¼ pumping rates at f1 and f2 Hz, bpd
H1, H2 ¼ developed heads at f1 and f2 Hz, ft
BHP1, BHP2 ¼ required brake horsepowers at f1 and f2 Hz, HP.
90
70 Hz
80
65 Hz BEP
70 Points
Developed Head, ft
60 Hz
60
55 Hz
50
50 Hz
40
45 Hz
30 40 Hz
20
10
0 1,000 2,000 3,000 4,000 5,000 6,000
Pump Capacity, bpd
Fig. 4.27 Head performance curves of an SN3600 pump at different frequencies.
The formulas given previously enable one to construct pump performance

curves for any electric frequency if the standard performance curve at
f ¼ 60 Hz is known. Figure 4.27 displays the head performance curves
of an SN3600 submersible pump at different frequencies. The three dashed
lines represent the loci of:
1. the lower limits of the recommended liquid rates,
2. the best efficiency points, and
3. the upper limits of the recommended liquid rates.
If this pump is driven by a 60 Hz electric supply, its recommended range is
between the rates of 2,400 and 4,600 bpd. It can easily be seen that by reg-
ulating the electric frequency from 40 Hz to 70 Hz, the pump can cover a
much wider range of flow rates, approximately from 1,600 bpd to
5,400 bpd. This extended operational range gives the much needed flexi-
bility for the designer/operator and makes it possible to easily compensate
for any uncertainties or changes that may occur in the well’s inflow
parameters.
The efficiency of the same pump driven at different speeds is shown in
Fig. 4.28. If compared to the constant frequency (60 Hz) operation, much
higher efficiencies can be attained at lower or higher liquid rates by chang-
ing the driving frequency. The pump’s optimum efficiency (values close to
170 Gabor Takacs
75
70
65
60
55
Pump Efficiency, %
50
45 Frequency (Hz)= 40 45 50 55 60 65 70
40
35
30
25
20
15
10
5
0
0 1,000 2,000 3,000 4,000 5,000 6,000
Pump Capacity, bpd
Fig. 4.28 Pump efficiencies of an SN3600 pump at different frequencies.
the BEP) can thus be maintained over the whole extended operational
range. This means lower energy requirements for those pumping rates out-
side the 60 Hz recommended range.
There are two solutions available for using variable frequency power in
ESP systems:
• variable speed drives (VSDs) utilize standard three-phase electric
power from a utility source and control their output frequency elec-
tronically, and
• variable frequency generators generate electric power on the wellsite
using some kind of engine and they control the output frequency by
regulating the speed of the engine.
4.6.2 Variable Speed Drives
Variable speed drives have many advantageous features and have been used
on an always increasing number of wells since their introduction in 1977
[41]. The different available VSD types and their operational features will
be described in this section.
In oil fields with ESP installations, electrical power supply may come:
• directly from a utility company,
• from the field’s power generators, or
• from a dedicated generator at the wellsite.
In either case, primary supply voltage is usually quite high and the required
surface voltages should be individually adjusted on each well. If a VSD unit
is also used, the general arrangement at the wellsite follows the schematic
Power Step-Down Step-Up ESP

VSD
Supply Transformer Transformer Motor
High Voltage 480 V 480 V Required V

60 Hz 60 Hz Required Hz Required Hz
Fig. 4.29 Typical electric power arrangement of an ESP well.
presented in Fig. 4.29. In this power flow, the VSD unit provides the
required frequency and step-down and step-up transformers ensure that
the required voltage is available at the wellhead.
4.6.2.1 Constructional Details
In order to optimize the operating conditions of electric motors, the use of
AC power supplies with variable frequencies is a widely used method in
many industries. Such power supplies come in two versions: voltage or cur-
rent source inverters. VSIs (voltage source inverters) are commonly used
in the ESP industry; they control the VSD’s output voltage and the current
fluctuates according to the load on the unit.
The VSD unit’s task is to convert the input frequency (usually 60 Hz) into
any frequency in its operating range. All VSDs [42] contain the following
three basic components (see Fig. 4.30):
1. the rectifier section converts the 60 Hz AC voltage and current into
a DC voltage and current,
2. the DC control section provides a smooth DC waveform to the
next section, and
3. the inverter section converts the DC voltage back to an AC voltage
at a selected frequency.
4.6.2.1.1 The Rectifier The simplest rectifiers use diodes to convert AC

to DC by allowing current flow in one direction only. The originally used
Fixed Freq. Variable Freq.

3 Phase AC 3 Phase AC
INPUT OUTPUT
Rectifier DC Control Inverter

Section Section Section
Fig. 4.30 Main components of a VSD unit.
172 Gabor Takacs
+ DC Output Voltage
Phase One
Voltage
Input
0 60 120 180 240 300 360
−
Phase Angle
Fig. 4.31 Output DC waveform from a “six-pulse” rectifier.
six-diode bridge configuration (see Fig. 4.30) creates a pulsating DC output

presented in Fig. 4.31. As seen, the DC voltage has six pulses for a complete
cycle, hence the name of this rectifier type: six-pulse. There are rectifiers
producing more pulses like 12, 18, 24, and so on, and the higher the pulse
number, the smoother the DC output. Smoother DC waveforms have the
added advantage of minimizing the amount of harmonics—that is, devia-
tions from a pure sine wave. Thus the average total harmonic distortion of
30% for the “six-pulse” rectifier goes down to about 8% for the 12-pulse
rectifier.
More advanced rectifiers use power electronic devices (SCRs ¼ silicone
controlled rectifiers, GTOs ¼ gate turn off thyristors, or IGBTs ¼ insulated
gate bipolar transistors), which allow a much better control of the output DC
waveform.
4.6.2.1.2 The DC Control Section The function of this section (also

called the DC bus) is to make the DC waveform as smooth as possible
so that the next section (the inverter) receives a smooth DC voltage.
The components used for this purpose are capacitors and inductors. Capa-
citors store energy and smooth out the pulses in the DC waveform when-
ever the rectifier voltage drops. Inductors, on the other hand, restrict the
amount of current that charges the capacitors and thus increase the dura-
tion of current flow. The effect of large enough inductors can even make
the current continuous thereby effectively reducing the amount of
harmonic distortions.
4.6.2.1.3 The Inverter The inverter converts the DC voltage produced by

the control section into an AC voltage at the selected frequency. The
inverter contains switchable power electronic devices (SCRs, GTOs,
IGBTs) to create a three-phase AC waveform. Each of these devices contains
a controllable switch and the output AC frequency is adjusted by the speed of

switching.
4.6.2.2 Available VSD Types
4.6.2.2.1 “Six-step” VSD The first VSD units manufactured in the late
1970s used Darlington transistors to create the AC waveform and produced
a very crude approximation of a sine wave. They control the output
frequency only and the voltage necessary for the ESP unit has to be reached
by the use of step-up transformers. The output waveform depends on the
configuration of the transformer, as shown in Fig. 4.32, and the drive is
named for the characteristic “six-step” shape of the △–Y connection.
Although used for more than 25 years, “six-step” VSDs are almost
completely replaced by more advanced units. Their main advantages are
the low investment and maintenance costs; the biggest disadvantage is
the very crude AC voltage waveform. As shown in Fig. 4.33, the output
voltage is only a crude approximation of a sine wave, the current being
more representative but it contains a large amount of harmonics.
4.6.2.2.2 Pulsed Width Modulation VSDs operating on the pulsed

width modulation (PWM) principle are widely used in other industries
because of their lower costs. They use inexpensive rectifiers with diodes
and inverters with IBGTs (insulated gate bipolar transistors). The output
voltage of these VSDs takes the form of a series of voltage pulses, each
of the same magnitude but having a different, controlled width. The out-
put frequency is controlled by adjusting the number of pulses per cycle
(called the “carrier frequency”) and the output voltage is controlled by
the width of the pulses.
Fig. 4.32 Voltage outputs from a “six-step”

Transformer Connected ∆ - ∆
Voltage
VSD unit.
Voltage
Transformer Connected ∆ - Y
174 Gabor Takacs
Fig. 4.33 Waveforms of a “six-step” VSD

unit.
Voltage
Current
Figure 4.34 shows two cases with an identical carrier frequency and thus
with the same output frequency but with different maximum output voltages
on the sine waveforms. Thinner pulses (upper part) result in a sine waveform
with a lower maximum value, whereas wider pulses (lower part) give higher
maximum voltage output. The ratio of the sine wave’s maximum voltage
and the voltage of the generated pulses is called the “modulation,” which is
about 50% in the first and almost 100% in the second case.
The output voltage and current of a typical PWM unit are shown in
Fig. 4.35 showing the characteristic chopped-up voltage waveform and
the quite smooth current that very closely approximates a perfect sine
wave. If compared to the output of older VSDs (see Fig. 4.33), units
Fig. 4.34 Voltage output at different 50% Modulation

modulations.
Voltage
95% Modulation
Voltage
Fig. 4.35 Outputs from a PWM type VSD.
Voltage
Current
working on the PWM principle provide much better waveforms, but they
have many disadvantages when used with cables longer than about 100 ft.
This is why a sine-wave filter is usually added to filter out harmonics and
to prevent other electrical problems.
4.6.2.2.3 Sine Wave Generators All electrical problems caused by har-

monics are eliminated if the waveform of the VSD’s output voltage is a pure
sine wave. As shown in Fig. 4.36, such units produce near-perfect voltage
and current waveforms [43] providing optimum conditions to the power
source and the ESP motor at the same time. The VSD units manufactured
according to the patented variable sine wave generation principle [44]
Fig. 4.36 Outputs from a sine wave generator.
Voltage
Current
176 Gabor Takacs
incorporate electronic components performing functions similar to a sine

wave filter and produce a cleaner sine wave without the use of an additional
filter.
The use of one leading manufacturer’s “Vector III” unit [45] increases
the quality of the sine wave by using rectifiers with 12 or 18 pulses. The
almost perfect sine wave output of the unit ensures the following
advantages:
• the efficiency of the VSD unit is improved to over 98%,
• output harmonics are reduced to less then 30% of those available from
“six-pulse” units, and
• motor and cable life is extended due to the lower heat loading effect of
the perfect sine wave.
4.6.2.3 Operational Characteristics
The AC electric power available from utility companies takes the form of
an almost perfect sine wave. This is also the waveform required for achiev-
ing optimum operating conditions for induction motors used in ESP units.
If a VSD unit is inserted between the power source and the ESP motor
then it is inevitable that the VSD unit’s presence affects the operation of
both the power source and the ESP unit. Since the DC control section
isolates the input and output sides of the VSD, the rectifier affects the
power source side only while the inverter affects the output side only.
Both effects are mainly due to harmonic distortion—that is, deviation of
the actual waveforms from a pure sine wave.
Harmonics are voltage or current loads which occur due to the devia-
tion from a pure sine wave, and have frequencies that are integral multiples
of the fundamental frequency. In a 60 Hz system these may be 120 Hz,
180 Hz, and so on, and are termed the 2nd, 3rd, and so on order harmo-
nics. The combined amount of harmonics is measured by the total har-
monic distortion (THD) which is defined as the ratio of the sums of the
total harmonic amplitudes and the fundamental amplitude in a cycle.
The harmonics affecting the power source are produced by the VSD’s
rectifier that constantly turns the diodes on and off to rectify the AC power.
The amount of harmonics can be reduced by rectifiers using more pulses
than the “six-pulse” rectifier. In this case smoother DC waveforms than
those given in Fig. 4.31 are produced. The most adverse effects of these har-
monics on the power system are (a) low power factors, (b) increased load and
(c) overheating of cables and other equipment. Of the VSD types, “six-step”
units have the worst THD values in the range of 25–70%, while sine wave
generators produce a THD of almost zero [46].
On the output side of the VSD unit, the converter produces the harmo-
nics reaching the ESP system. Harmonic distortion of the electric voltage
and current has different effects:
• High THD values of the voltage waveform mainly affect the electric
cable.
• High THD of the current results in:
– overheating of the motor,
– higher currents, and
– lower motor speeds.
“Six-step” units, again, generate high THD values and the best solution, as
before, is the use of sine wave generators.
4.6.3 Variable Frequency Generators

In oil fields with no or unreliable electrical power supply and/or in mobile
applications the use of electric generators on the wellsite is common. A newly
developed system [47] called the variable frequency generator (VFG) provides
a cost-effective supply of variable frequency power to ESP installations.
The variable frequency generator (VFG) consists of the following major
parts:
• a diesel engine providing the necessary continuous mechanical power
in a wide operational speed range,
• a three-phase AC generator with a widely varying output voltage and
frequency range, and
• a special electronic motor controller.
The unit’s power source is the diesel engine that supplies mechanical power
at variable speeds regulated by the VFG unit’s controller. The engine is
directly connected to the AC generator and turns the generator’s rotor.
Three-phase electricity is induced in the generator’s stator, the frequency
of which is proportional to the diesel engine’s speed. The generated three-
phase electric power is used to drive the ESP motor. All components of
the VFG unit are mounted on a self-contained skid unit and require signi-
ficantly less space than a conventional VSD system, thanks to the elimination
of several bulky parts like transformers, filters, and so on.
One of the greatest advantages of the described system lies in its output
waveform which is a perfect sine wave. Thus higher harmonics inevitably
produced by the operation of variable speed drives (VSDs) are completely
eliminated at any output frequency or power. Harmonics, as detailed pre-
viously, are responsible for power losses as well as for overheating of the
ESP motor and reduce the operating life of the motor lead extension,
178 Gabor Takacs
the electric cable, and so on. Pure sinusoidal power, in contrast, provides
ideal operating conditions to the motor and other components and may
significantly extend their run lives.
A recent case study [48] reports a motor current reduction of 4.5%, as
compared to the application of a VSD unit in a well producing about
19,000 bpd liquid. The VFG unit had a frequency range from 30 Hz to
60 Hz and was driven by an 855 HP diesel engine running at speeds
between 900 and 1,800 RPM.
4.6.4 Interaction of VSD/VFG and ESP Units
Before the discussion of the interaction of variable frequency power units
with the ESP system, the main features of ESP motors in a variable AC
frequency environment have to be presented. When an induction motor
is supplied with variable frequency at a constant voltage then the magnetic
flux density inside the motor changes and so does the torque developed. In
order to keep the torque constant, the ratio of the voltage and the fre-
quency has to be kept constant because motor torque equals voltage per
frequency squared. This is the reason why all VSD units’ output frequen-
cies and voltages satisfy the following equation:

f2
U2 ¼ U1 ð4:34Þ
f1

U1, U2 ¼ output voltages at f1 and f2 Hz, V.
The constant volts/frequency ratio ensures constant magnetic flux satura-
tion in the electric motor and the ESP motor becomes a constant-torque,
variable-speed device. As discussed in Chapter 2, motor speed directly
relates to the input frequency; and it follows from the constant ratio of
the input voltage and frequency that motor power is also linearly propor-
tional to input frequency [49, 50]. These relationships can be expressed as
given in the following:

f2
N2 ¼ N1 ð4:35Þ
f1

f2
HP2 ¼ HP1 ð4:36Þ
F1

N1, N2 ¼ motor speeds at f1 and f2 Hz, RPM
HP1, HP2 ¼ motor powers available at f1 and f2 Hz, HP.
If the power developed by an ESP motor and the power required by the
centrifugal pump, both valid at different frequencies, are compared, then it
is easy to see that motor power is a linear, while the required pump power
is a cubic, function of the driving frequency. (Compare Eqs. 4.36 and
4.33.) This means that the ESP pump and the motor, if operated at variable
electric frequencies, do not properly match each other [49–51].
The cooperation of an ESP pump and an electric motor at variable fre-
quencies is illustrated by the schematic drawing presented in Fig. 4.37. Since
motor power changes linearly, while pump power according to a cubic func-
tion (both as functions of the AC frequency), the motor is overloaded above
a given frequency ( fmax) and cannot be used with the given pump [52]. The
allowed maximum frequency can be found from the following formula:
rffiffiffiffiffiffiffiffiffiffiffiffi
HP1
fmax ¼ f1 ð4:37Þ
BHP1
150
140
130
120
110
Motor Power Motor
100
Load
Motor Load, %
90
Powers. HP
80
70 HPmotor
60
50
40 Pump
30 Power
BHPpump
20
10
0
60 Hz fmax
AC Frequency, Hz
Fig. 4.37 Pump and motor powers at variable frequencies.
180 Gabor Takacs
Alternatively, if the required pump brake horsepower, BHP1, at a given

frequency, usually at f1 ¼ 60 Hz, is known along with the maximum
desired frequency, fmax, then the minimum motor rating at 60 Hz should
be chosen as:
2
fmax
HP60 ¼ BHP60 ð4:38Þ
60
where: f1, fmax ¼ AC frequencies, Hz

BHP1, BHP60 ¼ required pump powers at f1 and 60 Hz, HP
HP1, HP60 ¼ motor nameplate powers at f1 and 60 Hz, HP.
The loading of the electric motor is shown in a dashed line in Fig. 4.37
and can be calculated at any frequency from the available motor power
and the required pump power, both at 60 Hz, as follows:

BHP60 f 2
Load ¼ 100 ð4:39Þ
HP60 60
As shown in Fig. 4.37, the motor is very lightly loaded at low frequencies,
a very unfavorable condition for induction motors whose efficiencies
diminish under light loads. This is the reason why, in order to ensure
favorable motor efficiencies, the operational frequency range should be
carefully selected.
When checking the strength of the pump shaft, the actual driving
frequency has to be considered and the shaft rating, given by the manu-
facturer at 60 Hz operation, must be corrected as given here:

f
HPshaft ¼ HPshaft60 ð4:40Þ
60
Alternatively, the maximum allowed frequency, fmax, which does not

overload the pump shaft, is found from:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
HPshaft60
fmax ¼ 60 ð4:41Þ
BHP60
where: BHP60 ¼ required pump power at 60 Hz, HP

HPshaft, HPshaft60 ¼ pump shaft rating at f and 60 Hz, HP.

Example 4.5
Select the proper ESP motor for an installation using 100 stages of an SN3600 pump
running at a frequency of 49 Hz and producing 3,000 bpd of liquid. At this frequency
and the required liquid rate the SN3600 pump develops 33 ft/stage and requires 1.1
BHP/stage.

Solution
The 100-stage pump requires 100 1.1 ¼ 110 HP at 49 Hz, which cor-
rected to 60 Hz, using Eq. 4.33, equals:
BHP60 ¼ 110ð60=49Þ3 ¼ 202 HP
Assume that a rotary gas separator is also required and it needs 5 HP

at 60 Hz, as found from manufacturer’s data. Since the separator is also a
centrifugal device, its power requirement has to be corrected the same
way:
BHPsep49 ¼ 5ð49=60Þ3 ¼ 3 HP
The total power requirement for pumping at the operational frequency of

49 Hz, therefore, equals 110 þ 3 ¼ 113 HP.
Assuming that a motor with this power is used, its required power at
60 Hz is found from Eq. 4.36 describing the performance of the ESP
motor:
HP60 ¼ 113ð60=49Þ ¼ 139 HP
Using Table D.2 from Appendix D, a Dominator 562 motor is selected

with the following parameters:
Power ¼ 150 HP
Voltage ¼ 3,190 volts
Amperage ¼ 28.5 amps
For checking the maximum allowed frequency, Eq. 4.37 is used:
fmax ¼ 60ð150=202Þ0:5 ¼ 52 Hz
182 Gabor Takacs
250
200
Motor Load, %
150
Motor Power
Power, HP
100
Motor Load
50
Pump
Power
0
0 10 20 30 40 50 60 70
AC Frequency, Hz
Fig. 4.38 Variation of motor and pump powers for Example 4.5.
Note that although this value is greater than the required 49 Hz, the VSD unit
should never be run above this frequency because the motor then will be over-
loaded. Choosing a bigger capacity motor would eliminate this problem.
Actual loading of the motor is found from Eq. 4.39 where the power
required for driving the separator is also included:
Load ¼ 100 207=150ð49=60Þ2 ¼ 92%
The pump shaft strength is checked next. Using the shaft limit of 256 HP,
found from Table C.1 in Appendix C, the shaft capacity at 49 Hz is
found from Eq. 4.40:
HPshaft ¼ 256ð49=60Þ ¼ 209 HP; greater than the required pump HP
Finally, the motor voltage required to achieve full saturation at 49 Hz is

calculated from Eq. 4.34:
Umotor ¼ 3; 190ð49=60Þ ¼ 2; 605 volts

Calculation results, including the motor loading, are displayed in Fig. 4.38
in the function of AC frequency.
4.6.5 Benefits of Using VSD/VFG Units

If compared to conventional ESP installations with constant motor speeds,
installations running at variable frequencies have several advantages. The
most important general improvement is the wide flexibility of the variable
frequency ESP system that permits perfect matching of the lift capacity of
the ESP system and the well’s productivity. This means that, in contrast to
fixed-frequency units, great uncertainties or extensive changes in well
inflow conditions can easily be accommodated during installation design
or later. Thus an installation designed with unreliable well data must not
be choked or cycled to produce the desired liquid rate. Choking involves
wasting of electric power, while cycling, on the other hand, decreases the
run life of the ESP equipment. With a variable frequency unit, these
problems are eliminated.
Additional benefits of both kinds of variable frequency units (VSDs and
VFGs) can be summarized in the following [53–56].
• Liquid rates from wells with small-diameter casing strings can be
increased by running the ESP unit at higher speeds [50].
• The efficiency of the ESP system increases if a high capacity pump is
slowed down to produce a lower rate because the efficiency of larger
pumps is higher even at lower speeds.
• The life of ESP equipment is increased compared to that of an over-
loaded unit running at a constant speed because of the lower heating
effect and the lower thrust loads.
• Variable frequency units facilitate “soft starting” of the ESP installation
and reduce the startup power requirements compared to an “across-
the-line” start. This feature very efficiently reduces the startup currents
and extends the operational life of ESP motors.
• Premature ESP motor failures due to power system disturbances (volt-
age transients, lightning strikes, etc.) are eliminated because the ESP
motor is automatically isolated from the power source.
• Variable speed ESP units can perform well testing operations and help
select the optimum equipment for the given well.
• Closed-loop control of the installation is possible when motor fre-
quency is varied according to well inflow rate, sensed by downhole
instrumentation [55].
• Gas lock problems are eliminated if the pump is slowed down before a
complete gas lock develops.
184 Gabor Takacs
The advantages of variable frequency generators (VFGs) over VSD units

are given in the following:
• Investment costs can be lower because of the elimination of several
system components like transformers, filters, and so on.
• Maintenance requirements and costs of the VFG units can be lower
due to their simplicity.
• Because of the perfect sine waveform output, heat loading of the
motor and the cable is reduced increasing the reliability and run life
of ESP equipment.
• Under similar well conditions, motor current is reduced because no
harmonic distortion occurs.
• The VFG unit usually needs less space than a VSD unit; application in
offshore conditions is easier.
References
1. Tovar, C. A.: “Sizing and Selecting ESP for Heavy Crude Oil Applications.” Paper
presented at the ESP Workshop held in Houston, April 29–May 1, 1998.
2. Stepanoff, A. J.: Centrifugal and Axial Flow Pumps. Theory, Design, and Application. 2nd
Ed. John Wiley & Sons, Inc., New York, 1957.
3. Determination of Pump Performance when Handling Viscous Liquid. Hydraulic Institute
Standards, 20th Ed., 1969.
4. Turzo, Z., Takacs, G. and Zsuga, J.: “Equations Correct Centrifugal Pump Curves for
Viscosity.” OGJ, May 29, 2000, 57–61.
5. Submersible Pump Handbook. 6th Ed. Centrilift-Hughes, Inc. Claremore, Oklahoma,
1997.
6. Riling, G.: “Sizing Submersible Pumps for Viscosity Effects.” PE, July 1976, 64–72.
7. Information retrieved from www.centrilift.com.
8. Alhanati, F. J. S.: “Bottomhole Gas Separation Efficiency in Electrical Submersible
Pump Installations.” PhD Dissertation, University of Tulsa, 1993.
9. Standing, M.B.: “A Pressure-Volume-Temperature Correlation for Mixtures of
California Oils and Gases.” API Drilling and Production Practice, 1947, 275–86.
10. Papay, J.: “Change of Technological Parameters in Producing Gas Fields.” (in
Hungarian) Proc. OGIL Budapest. 1968, 267–73.
11. Hankinson, R.W., Thomas, L.K. and Phillips, K.A.: “Predict Natural Gas Properties.”
Hydrocarbon Processing, April 1969, 106–8.
12. Lea, J. F. and Bearden, J. L.: “Effect of Gaseous Fluids on Submersible Pump Perfor-
mance.” JPT, December 1982, 292–30.
13. Sachdeva, R., Doty, D. R. and Schmidt, Z.: “Performance of Electric Submersible
Pumps in Gassy Wells.” SPE Production and Facilities, February 1994, 55–60.
14. Pessoa, R. and Prado, M.: “Two-phase Flow Performance for Electrical Submersible
Pump Stages.” SPE Production and Facilities, February 2003, 13–27.
15. Sheth, K. and Bearden, J.: “Free Gas and a Submersible Centrifugal Pump: Application
Guidelines.” Paper presented at the ESP Workshop held in Houston, May 1–3, 1996.
16. Dunbar, C. E.: “Determination of Proper Type of Gas Separator.” SPE Microcom-
puter Users Group Meeting, Long Beach, California, October 15–17, 1989.
17. Lea, J. F., Turpin, J. L. and Bearden, J. L.: “Correlation of Performance Data for
Electric Submersible Pumps with Gas-Liquid Flow.” Proc. 33rd Southwestern
Petroleum Short Course, Lubbock, Texas, 1986, 267–81.
18. Wilson, B. L.: “There’s No Free Lunch: Pumping Two Phase Fluids with ESP.” Paper
presented at the ESP Workshop held in Houston, April 30–May 2, 2003.
19. Gas Handling Guideline. Baker Hughes Centrilift, Claremore, Oklahoma, January 2000.
20. Wilson, B. L.: “ESP Gas Separator’s Affect on Run Life.” Paper SPE 28526, 1994.
21. Wilson, B. L., Mack, J. and Forster, D.: “Operating Electrical Submersible Pumps
below the Perforations.” SPE Production and Facilities, May 1998, 141–5.
22. Bunnelle, P. R.: “Liquid-Gas Separator Unit.” US Patent 3,887,342, 1975.
23. Tuzson, J. J.: “Separator.” US Patent 4,088,469, 1978.
24. Lee et al.: “Liquid-Gas Separator Apparatus.” US Patent 4,481,020, 1984.
25. Kobylinski, L. S., Taylor, F. T. and Brienan, J. W.: “Development and Field Test
Results of an Efficient Downhole Centrifugal Gas Separator.” JPT, July 1985,
1295–304.
26. Alhanati, F. J. S., Sambangi, S. R., Doty, D. R. and Schmidt, Z.: “A Simple Model for
the Efficiency of Rotary Separators.” Paper SPE 28525 presented at the 69th Annual
Technical Conference and Exhibition held in New Orleans, September 35–38, 1994.
27. Harun, A. F., Prado, M. G. and Doty, D. R.: “Design Optimization of a Rotary Gas
Separator in ESP Systems.” Paper SPE 80890 presented at the Production and
Operations Symposium held in Oklahoma City, March 22–25, 2003.
28. Hall, O. P. and Dunbar, C. E.: “Computer Helps Select Best Lifting Equipment.”
OGJ, May 10, 1971, 84–8.
29. Swetnam, J. C. and Sackash, M. L.: “Performance Review of Tapered Submergible
Pumps in the Three Bar Field.” JPT, December 1978, 1781–7.
30. Wilson, B. L.: “Gas Handling Centrifugal Pumps.” Paper presented at the ESP
Workshop held in Houston, April 29–May 1, 1998.
31. Villamizar, J. J., Beltran, A., Guindi, R., Flores, G. F., Afanador, L. and Aguilar, H.:
“Performance Evaluation of a new Gas Handler in an Oil Well with more than 60%
Free Gas at the Intake.” Paper presented at the ESP Workshop held in Houston, April
30–May 2, 2003.
32. Hirth, B., Waygood, R., Justrom, P., Miller, J., Ossia, S. and Ollre, C.: “Case Study:
Performance Evaluation of a Helicoaxial Multiphase Pump in a CO2 Flood.” Paper
presented at the ESP Workshop held in Houston, April 27–29, 2005.
33. “Poseidon Multiphase Gas-handling System.” Schlumberger brochure, August 2003.
34. Information retrieved from www.centrilift.com.
35. Wilson, B. L.: “The Effects of Abrasives on Electrical Submersible Pumps.” SPE
Drilling Engineering, June 1990, 171–5.
36. Information retrieved from www.woodgroup-esp.com.
37. King, D. G., Traylor, F. T. and Stewart, R. E.: “Abrasion Technology for Electric
Submergible Pumps.” Paper SPE 12199 presented at the 58th Annual Technical
Conference and Exhibition held in San Francisco, October 5–8, 1983.
38. Sand Handler – The Centrilift Abrasion Resistant Pumping System. Centrilift Co., Clare-
more, Oklahoma, 1996.
39. Chilcoat, D. C.: “Bearing Selection for Specific Applications.” Paper presented at the
ESP Workshop held in Houston, April 28–30, 1999.
40. Fuller, D., Fickes, B. and Dowdy, R.: “Electric Submersible Pumping Systems Applied
in High Temperature Environments.” Paper presented at the ESP Workshop held in
Houston, April 27–29, 1994.
41. Divine, D. L.: “A Variable Speed Submersible Pumping System.” Paper SPE 8241
presented at the 54th Annual Technical Conference and Exhibition held in Las Vegas,
September 23–26, 1979.
42. Alcock, D. N.: “Application of Variable-frequency Drives to Deep-well Submersible
Pumps.” PEI, March 15, 1980, 36–44.
186 Gabor Takacs
43. Breit, S., Sikora, K. and Akerson, J.: “Overcoming the Previous Limitations of Vari-
able Speed Drives on Submersible Pump Applications.” Paper SPE 81131 presented at
the Latin American and Caribbean Petroleum Engineering Conference held in Port-
of-Spain, April 27–30, 2003.
44. LeDoux, K. and Pelton, W. C.: “Sine Wave Variable Speed Drive.” US Patent
6,531,842, 2003.
45. Vector III Brochure. Wood Group ESP Inc., 2001.
46. Harmonics Tutorial. Wood Group ESP Inc., 2001.
47. Information retrieved from www.canadianadvanced.com.
48. Pettigrew, D. R. and Limanowka, W. A.: “Variable Frequency Generator as the
Power Supply for an ESP.” Paper presented at the ESP Workshop held in Houston,
April 27–29, 2005.
49. Wilson, B. L. and Liu, J. C.: “Electrical Submersible Pump Performance using Variable
Speed Drives.” Paper SPE 13805 presented at the Production Operations Symposium
held in Oklahoma City, March 10–12, 1985.
50. Powers, M. L.: “Effects of Speed Variation on the Performance and Longevity of
Electric Submersible Pumps.” SPE PE, February 1987, 15–24.
51. Butlin, D.: “Variable-speed Controller Provides Flexibility to Electric Submersible
Pumps.” OGJ, June 9, 1986, 44–47.
52. “Recommended Practice for Sizing and Selection of Electric Submersible Pump
Installations.” API RP 11S4, 3rd Ed., American Petroleum Institute, 2002.
53. Divine, D. L.: “Variable Speed Submersible Pumps Find Wider Application.” OGJ,
June 11, 1979, 55–61.
54. Divine, D. L.: “Application and Economics of Variable Speed Submersibles.” WO,
June 1979, 133–8.
55. Alcock, D. N.: “Operating Submersible Pumps with Closed-loop Speed Control.”
OGJ, October 12, 1981, 115–28.
56. Leuthen, M. and Radcliff, T.: “Variable Speed Drives: Definitions, Applications and
Comparisons.” Paper presented at the ESP Workshop held in Houston, April 30–May 2,
1997.

Use of ESP Equipment in Special Conditions

Uploaded by

Copyright:

Available Formats

Use of ESP Equipment in Special Conditions

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Use of ESP Equipment in Special Conditions

Uploaded by

Copyright:

Available Formats

CHAPTER 4

Use of ESP Equipment in

Electrical Submersible Pumps Manual ISBN 978-1-85617-557-9

4.2 PUMPING VISCOUS LIQUIDS

conditions viscosity is not taken into consideration when selecting the

where: QwBEP ¼ water rate at BEP, 100 gpm

CQ ¼ 1:0 4:0327 103 Q 1:724 104 ðQ Þ2 ð4:6Þ

C ¼ 1:0 3:3075 102 Q þ 2:8875 104 ðQ Þ2 ð4:7Þ

CH0:6 ¼ 1:0 3:68 103 Q 4:36 105 ðQ Þ2 ð4:8Þ

CH0:8 ¼ 1:0 4:4723 103 Q 4:18 105 ðQ Þ2 ð4:9Þ

CH1:0 ¼ 1:0 7:00763 103 Q 1:41 105 ðQ Þ2 ð4:10Þ

CH1:2 ¼ 1:0 9:01 103 Q þ 1:31 105 ðQ Þ2 ð4:11Þ

where: Q ¼ pump capacity, bpd

Point Q bpd H ft Eff. %

QwBEP ¼ 900 42=1440 ¼ 26:25 gpm ¼ 0:2625 100 gpm

Now, using Eq. 4.4, Q is found:

Q∗ ¼ exp½ð39:5276 þ 26:5605 ln 88 þ 4:276Þ=51:6565 ¼ exp 3:15 ¼ 23:34

CQ ¼ 1 4:0327 103 23:34 1:724 104 23:342 ¼ 0:812

C ¼ 1 3:3075 102 23:34 þ 2:8875 104 23:342 ¼ 0:385

CH0:6 ¼ 1 3:68 103 23:34 4:36 105 23:342 ¼ 0:890

CH0:8 ¼ 1 4:4723 103 23:34 4:18 105 23:342 ¼ 0:873

CH1:0 ¼ 1 7:00763 103 23:34 þ 1:41 105 23:342 ¼ 0:844

CH1:2 ¼ 1 9:01 102 23:34 þ 1:31 104 23:342 ¼ 0:797

visc3 ¼ 0.3854 64.0 ¼ 24.7, and

BHPvisc ¼ CBHP BHPw gl ð4:13Þ

Table 4.1 Viscosity correction factors according to Centrilift [7].

Viscosity SSU Correction factors

50 1.000 1.000 0.945 1.058

Provided courtesy of Centrilift.

viscosity measured in SSU (Saybolt Seconds Universal) units. To convert

v ¼ 2:273½88 þ ð882 þ 158:4Þ0:5 ¼ 402 SSU

CQ ¼ 0:847; CH ¼ 0:909; C ¼ 0:497; CBHP ¼ 1:549:

4.3 PRODUCTION OF GASSY WELLS

being proportional to the density of the fluid pumped. Because of their

4.3.2 Free Gas Volume Calculations

q0g ¼ qo ðGOR Rs ÞBg ð4:15Þ

where: qg0 ¼ in-situ gas volumetric rate, ft3/d

where: vsl ¼ liquid superficial velocity, ft/sec,

where: ql ¼ liquid volumetric rate, STB/d

where: s ¼ interfacial tension, lb/sec2

where: gg ¼ specific gravity of the gas, –

where: go, gw ¼ specific gravities of the oil and water, –

where: qo ¼ oil volumetric rate, STB/d

The Standing bubblepoint pressure correlation [9] allows the calculation

where: y ¼ 0.00091 T 0.0125 API

where: PIP ¼ pump intake pressure, psia

Ppc ¼ 709:6 58:7 gg ð4:27Þ

Tpc ¼ 170:5 þ 307:3 gg ð4:28Þ

where: gg ¼ gas specific gravity, –.

Bo ¼ 0:972 þ 1:47  104 F 1:175 ð4:29Þ

where: F ¼ Rs(gg/go)0.5 þ 1.25 T

Oil API degree ¼ 30

y ¼ 0:00091 150 0:0125 30 ¼ 0:238

Rs ¼ 0:6½1; 000=ð18 100:238 Þ1:205 ¼ 147 scf =bbl

Ppc ¼ 709:6 58:7 0:6 ¼ 674:4 psi

The reduced state parameters are:

Ppr ¼ 1; 000=674:4 ¼ 1:483

Tpr ¼ ð150 þ 460Þ=354:9 ¼ 1:719

CQ ¼ 1:0 4:0327 103 Q 1:724 104 ðQ Þ2 ð4:6Þ

C ¼ 1:0 3:3075 102 Q þ 2:8875 104 ðQ Þ2 ð4:7Þ

CH0:6 ¼ 1:0 3:68 103 Q 4:36 105 ðQ Þ2 ð4:8Þ

CH0:8 ¼ 1:0 4:4723 103 Q 4:18 105 ðQ Þ2 ð4:9Þ

CH1:0 ¼ 1:0 7:00763 103 Q 1:41 105 ðQ Þ2 ð4:10Þ

CH1:2 ¼ 1:0 9:01 103 Q þ 1:31 105 ðQ Þ2 ð4:11Þ

Now, using Eq. 4.4, Q is found:

Bo ¼ 0:972 þ 1:47 104 F 1:175 ð4:29Þ