Pre- Field Trial Testing of a Twin Screw Multiphase Pump
Pre- Field Trial Testing of a Twin Screw Multiphase Pump
M. Yamashita, Y. Sharma, SPE, M. Ihara, SPE, Japan National Oil Corp. and K. Yamada, Kosaka Laboratory Limited
Abstract
Performance tests were conducted on a twin screw multiphase
pump in a large-scale flow loop prior to its first trial in an oil
field in the Middle East. The main objectives of these tests
were (1) confirmation of the pump’s hydrodynamic
performance by examining the flow rate-head (Q–H) curves
and the efficiency-head (η–H) plots, (2) checking the integrity
of the design and construction, and (3) obtaining operational
experience with the complete system. The pump was installed
at the exit of a 82 m long, 106.3 mm ID pipeline and upstream
of a three phase high pressure separator. The fluids used were
nitrogen and water. A total of 76 flow tests were executed
under the following experimental conditions; gas volume
fraction (GVF) at the suction: 0 – 95 %; pump speed: 1200 –
1800 revolutions/minute; and differential pressure: 142 – 350
psig. These ranges covered most of the operating conditions to
be expected during the field trial.
The pump operated successfully in the flow loop for a total
of 150 hours. The Q-H curves showed an increase in
volumetric efficiency with an increase in GVF. At high GVF,
above 80%, an almost critical flow situation was attained,
where the total flow rate was independent of the differential
pressure. The η–H plots showed that the pump efficiency
decreases with an increase in GVF. These characteristics are
consistent with the results of tests conducted by other
manufacturers on similar pumps. During the tests, no leaks
were obtained through the mechanical seals and there were no
problems with the quenching unit and cooling at the bearings.
A maximum rise in temperature of 13 °C occurred when the
GVF was 95%. Scratches seen on the external face of the
screw and the internal face of the sleeve were attributed to
scales in the pipeline. As a result of these data, the pre-field
trial tests on the twin screw pump were considered successful
and ready for testing in an oil field.
Introduction
Multiphase boosting adds energy to unprocessed fluids and
permits them to be transported through longer tie-backs to
processing facilities located downstream. A reduction or
elimination of the production infrastructure such as platforms
and separating equipment is therefore possible. This would
lead to lower operating costs associated with the development
of hydrocarbon reserves. In this way, marginal fields located
in hostile environments can be developed economically. A
multiphase pump can also reduce the back pressure on
producing formations, leading to an increase in production,
and possibly higher recovery factors.
Generally, two types of boosting systems used in
multiphase flow applications are (1) rotodynamic and (2)
positive displacement pumps. The main characteristics of
these pumps have been identified by Wietstock and Witte1. In
rotodynamic pumps, the differential pressure is a function of
the density of the fluid, pump speed, and pump construction. It
can yield high pressure differences if the multiphase mixture is
homogeneous, densities are high, and it is operated at high
speeds. A variable speed rotodynamic pump must be used to
maintain a constant pressure when the GVF changes at the
suction. They can deliver high discharge flow rates and have
the ability to handle fluids with a high solids content.
In the case of a positive displacement pump, the pressure
difference is independent of the density of the fluid. It is
dependent on the pump’s speed and geometry. It has the
advantage over the rotodynamic type in being able to handle
flows with varying inlet GVF’s with a constant speed. Also, it
can be applied to low density fluids. It has disadvantages of
low discharge rates, complicated mechanisms, and low solids
handling capability.
Some of the challenges, which have confronted the
development of the twin-screw multiphase flow boosting
technology, have been identified since its early days by
Rohlfing and Brandt2. They include; (1) gas handling
capability, (2) control of compression heat, and (3) efficiency.
In addition to these flow issues, safety considerations such as
control of back spin, safety valve arrangement, discharge
pressure control, and speed control have been noted.
Challenges also exist in the development of mechanical seals,
seal oil systems, and lube oil systems.
This paper focuses on the twin screw positive
displacement pumping system. The pump was developed as
part of a Joint Industry Project (JIP) on multiphase flow
boosting. The paper addresses the pre-field trial tests of the
pump conducted at a large-scale multiphase flow test facility.
Description of Twin Screw Pump
The pump developed in the current JIP was designed based on
existing models of conventional twin screw pumps as shown
in Fig. 1 and the results of two-phase (air-water) flow
experiments3.
Two parallel shafts with intermeshing screws operate
within close fitting bores. Each shaft has right hand and left
hand screws as integral parts. Contact of the screws is avoided
with the use of externally mounted bearings and timing gears.
Advantages of this configuration have been identified by
Prang4. The right hand and left hand screws on each shaft (1)
doubles pump capacity, and (2) reduces axial hydraulic
thrusts. The integral design of the screw and shaft reduces
shaft deflection and wear, thereby providing a longer life.
Fluids enter the bore and the spaces between the screws.
The rotating action of the screws squeezes the fluids from one
screw to the other. In this way, the fluid is advanced from the
suction to the discharge. Slippage results in displacement less
than the theoretical. It depends on clearance, viscosity,
differential pressure, and the number of trapped areas of fluid
referred to as locks. In his paper on selecting multiphase
pumps, Prang3 identified the main parameters to consider in
designing twin screw pumps. They include, (1) pitch, (2)
screw length, (3) clearances, (4) screw profiles, (5) shaft
diameters, (6) casing style, (7) sealing methods, and (8)
sealing materials.
A summary of the specifications of the pump used in the
trial is as follows:
- External Dimensions (with motor and skid) :
L=3128 mm, W=1100 mm, H=1715 mm
- Screw Dimensions and Arrangement :
external diameter of screw : 193 mm
external diameter of shaft : 127 mm
pitch of thread : 24 mm
lead (length) of screw : 171 mm
number of lock : 6
flow direction : end suctions / center discharge
- Weight (with motor and skid) : 4700 kg
- Design Pressure : 800.6 psig
- Normal Suction Pressure : 100 psig
Normal Discharge Pressure : 450 psig
- Theoretical Capacity : 73.9 m3/h @1800 rev/min
- Actual Capacity :
48 m3/h @GVF 0%, 1800 rev/min, ΔP= 350 psi
62 m3/h @GVF 95%, 1800 rev/min, ΔP= 350 psi
- GVF Range: 0% to 95%
- Seal: 4 double mechanical seals (oil seal type)
- Auxiliaries: Quenching Unit, Inverter Panel, and Control
Panel
The number of screw threads was optimized by previous
experiments using a conventional twin-screw pump. The flow
direction was determined considering seal systems. In order to
minimize the load on seals, the suctions were placed at both
ends of screw. Also, the length of the screws was minimized
to reduce the impact of bending on the gaps between screws.
The pump was installed at the exit of a 82 m long, 106.3
mm ID pipeline and 21 m upstream of a three phase high
pressure separator. Details of the experimental facilities are
described in the following section.
Experimental Facilities
The major specifications of the multiphase flow test loop at
the Kashiwazaki Test Field of the Japan National Oil
Corporation, include,
- Test Fluids : Gas Phase = Air or Nitrogen,
Liquid Phase = Water and Kerosene
- Test Pipelines : 500m-long 106.3 mm diameter and 200mlong
54 mm diameter
- Fluid Injection Points: approx. 100 m, 200 m, and
500 m upstream from the separator
- Maximum Operating Pressure : 493 psig
- Inclined Section : 10 m long, 0 – 90 degrees
- Hilly-terrain Section: 4 × 10 m slopes, –10 to 10 degrees
- Instrumentation: Sensors for pressure, temperature, and liquid
holdup
- Supply System :
Water Pump : 180m3/h × 493 psig
Kerosene Pump : 180m3/h × 130 psig and 493 psig
Gas Compressor : 2000 Nm3/h × 130 psig and 493 psig
Separator : 2000 Nm3/h (gas), 360 m3/h (liquid)
Water Tank : 27.5 m3, Kerosene Tank : 33 m3 .
Details of the flow loop are shown in Fig. 2.
Flow Tests
The flow tests were conducted with the pump and its
quenching unit installed at the end of the 4 in. pipe and
upstream of the horizontal separator as shown on Fig. 3.
Nitrogen was used as the gas phase and water as the liquid
phase. The fluids were injected 100 m upstream of the
horizontal separator which was kept at a constant pressure
around 427 psi. The testing procedure can be summarized as
follows:
1. The revolution speed of the pump was set at 1200, 1500,
or 1800 rev/min.
2. A targeted GVF was set by adjusting the liquid and gas
flow rates at the injection point. These rates are maintained
steady during the test with the use of flow controllers.
3. The suction and discharge pressures of the pump were
measured after stabilization. The discharge pressure of the
pump was controlled around 427 psig by a pressure control
valve of the separator.
Consequently, the data point (liquid flow rate, differential
pressure) was obtained for the GVF. The steps 1. to 3. were
repeated to obtain the data necessary to construct the Q–H
curves.
A total of 76 operating conditions was tested to obtain the
pump hydrodynamic performance. These conditions included
the following;
- GVF : 0, 20, 40, 60, 80, 90, and 95%
- Pump speed: 1800, 1500, and 1200 rev/min
- Differential pressure: ΔP:142, 213, 284, and 350 psi.
Fig. 4
presents a summary of the tests on the flow pattern map
of Taitel and Dukler5.
The mechanical condition of components was also
monitored. These checks included the behavior of the
mechanical seals and the capacity of cooling system. The
inverter and control panels were not tested as they were
specially designed for the field test.
Results and Discussion
Operability.
The pump operated for a total of 150 hours
during the pre-field trial tests. The operation in a closed loop
was executed successfully according to the procedure
described above.
Q-H Performance. Fig. 5
shows the hydrodynamic pump
performance curves. Each curve represents a plot of liquid
capacity versus the differential pressure (head) for a fixed
GVF and pump speed. It should be stated that the performance
of the pump with single phase water only (GVF 0%) was the
same as the manufacture’s factory test. The figure shows that
at high GVF conditions, the pump's capacity for the liquid
phase was not significantly affected by the differential
pressure. Under these conditions, the GVF and the speed
determined its capacity. This was not the case at low GVF's,
where the differential pressure was also important.
This phenomenon described above is similar to that of
critical flow. An explanation of this is given here, and follows
that of Prang4. The difference between the volume transferred
by the screw cavities (locks) and the back flow (slippage)
through gaps between the screws and the sleeve determine the
pump's capacity. The theoretical displacement of the pump is
determined by the pump's speed and the size of the screw.
Slippage through finite clearances to the suction area result in
the actual capacity being less than the theoretical value. In
single-phase incompressible flow, the pressure distribution
from suction to discharge along the screw increases
proportionally to the transferred distance. Previous work6 has
indicated a steep pressure gradient near the discharge on the
screw for two-phase gas-liquid flow as shown in Fig. 6. Gas is
compressed toward the discharge end while the capacity of the
screw cavity is constant. Slippage near the discharge is higher
than near the suction, resulting in a higher pressure gradient.
This effect was more pronounced with higher GVF because of
increased compressibility. Such a phenomenon may produce
the condition like critical flow. The slip at the last lock causes
compression of the fluid in the screws. The slippage may not
extend to the suction area if there are a sufficient number of
locks. As a result, the amount of fluid drawn at the suction is
independent of the differential pressure.
The performance curves for total capacity are shown in
Fig. 7
using the volumetric efficiency at the suction. This
efficiency is defined as the ratio of the total volume of injected
gas and water to the theoretical capacity. In the case of a pump
speed of 1800 rev/min, higher GVF's resulted in higher
volumetric efficiencies up to a GVF of 90%. A reverse in this
behavior was observed for GVF's greater than 90% and at low
differential pressures. In the case of a pump speed of 1200
rev/min, a similar reverse of the volumetric efficiency was
found out between 60% and 80% in GVF. The reduction in the
efficiency at high GVF's has been attributed to gas locking at
the screws2. It should be noted from Fig.4 that the flow pattern
changes from slug to stratified / wavy flow as the GVF
increases from 80% to 95%. This flow pattern change may
also result in a change in the trend of the pressure gradient
curve (see Fig. 6), and hence a change in the suction-discharge
behavior. Also, as shown in Fig. 7, higher differential
pressures improve the efficiency, for both pump speeds in
high GVF condition. Higher differential pressure means lower
suction pressure in this trial like as practical operations, while
most of flow loop tests for this type of pumps had been
usually conducted in fixed suction pressures. Low suction
pressure has the same effect as the increase of GVF7. These
observations, changing flow patterns, and operating the pump
at low and high differential pressures on the pump's
performance require further investigation.
Fig. 7
also shows higher volumetric efficiency with higher
pump speed. It can be also described by Fig. 8. Higher
revolution speeds result in lower slippage in gas-liquid two
phase flow, while the speed does not effect the slippage in
single phase liquid flow.
η-H Performance. Fig. 9 show the hydrodynamic pump
efficiency curves; pump efficiency, η versus the head obtained
in the pre-trial tests. The kinetic energy of fluids at the suction
and the discharge was predicted to be much smaller than the
potential energy. As a result, the kinetic energy was ignored in
this analysis. In order to estimate the kinetic energy effects,
measurement of liquid hold-up would be required. This value
was not measured in the experiments. It should be noted that
the result at GVF 0% was the same as that of the
manufacture’s factory test.
The figures show that at a given pump speed, the pump's
efficiency decreases with increasing GVF. This can be
explained with the aid of Fig. 7. For a given total capacity at
the inlet, a higher GVF requires a higher differential pressure
and hence a higher horsepower. This should not be the case,
since we would expect the lighter (higher GVF) fluid to
require less power. Fig . 9 also show that the efficiency of the
pump increases with pump speed.
The screw axis of a twin-screw pump bends by gravity and
reaction forces to fluid pressure. Thus, the dimension of gaps
should be optimized considering such bending. Moreover, the
gaps of this type of pump, for oil production, should be
constructed larger than the designed conditions. This is due to
the possibility that the suction pressure of these pumps may
exceed its discharge pressure at times, e.g. under its start-up
and shutdown. This would cause the axis to bend in a contrary
manner to regular operation. This factor was taken into
consideration in the construction of the pump under
investigation.
Confirmation for Mechanical Property.
Seal System.
No leakage through mechanical seals was
observed during the pre-field trial experiments. Four
mechanical seals of the pump were removed and investigated
to determine the degree of damage after running for 150 hours.
Small scratches were found on the internal face of the seal ring
placed on the non-drive end of the drive shaft. It was assumed
that solids in the seal fluid caused this damage. The external
ring and other three seals had zero damage.
The seal fluid pressure was kept higher than the internal
fluid pressure during the pump's operation. A decrease of
stored seal oil was not observed.
Quenching Unit for Seal System.
The performance of this
unit was confirmed without any problems. The unit had an
automatic pressure control valve to govern the seal fluid
pressure. It was set manually because its controller was
installed in the control panel designed specifically for the
field. The control valve was fixed at 43 psi higher than the
suction pressure of the pump under steady conditions. It was
reset to its highest set point (700 psi) during start-up, shutdown,
and changing the flow conditions.
Cooling.
Each bearing of the pump and the quenching unit
required cooling water. The cooling water flow rate of the test
loop was less than half of the recommended value. However,
there was no problem with overheating, since the temperatures
of the water and atmosphere were lower than designed values.
Temperature Increase.
The temperature of the waternitrogen
mixture rises by compression as it passes through the
pump. This effect is enhanced at higher GVF's. This
temperature increase should be minimized for safe operations.
A large difference of temperature influences the dimension of
the screw and gaps. The recommended range of the
differential for the pump was 15 °C. Fig. 10 shows the result
of the differential temperature during the pre-field trial tests.
The highest value of 13 °C was recorded with the condition of
GVF 95%. It was small when the GVF was less than 80%.
During the actual field trial, higher differential temperatures
are expected due to the higher temperatures of the atmosphere
and the produced hydrocarbon fluids, and lower heat
transportability because of lower specific heats. This is
anticipated, even though heating up by adiabatic compression
of a hydrocarbon gas is less than nitrogen.
Screw Damage.
Scratches were found on the external face
of screw and internal face of the sleeve. During the pre-field
trial tests, a strainer placed at the pump's suction was broken
by the high differential pressure. The pump immediately
tripped by an over current detection in the existing inverter.
Many scales in the pipe of the test loop might have caused this
damage. It is also possible that the damage to the screw
mentioned above might be caused by the fragments of the
strainer element. The pump was restarted after a hand turning
check and reinstallation of the strainer. No capacity reduction
was observed.
Conclusions
1. The hydrodynamic performance of the twin screw pump
developed in a JIP was investigated.
2. The volumetric efficiency was shown to be dependent on
the GVF, differential pressure, and pump speed. It is
suggested that further investigation be done on the effect of
changing flow patterns, and operating the pump at low and
high differential pressures on the pump's performance.
3. The tests confirmed the pump's operation procedure in
the closed loop.
4. The seal system had no problem. It is important to
remove contaminants around the mechanical seals and in the
quenching fluid. The seal fluid pressure should be kept higher
than the suction pressure of the main pump during its
operation. The cooling system was not thoroughly checked
during the tests. It should be carefully monitored during
operations in the field.
5. A maximum rise in temperature of 13 °C was observed
during the tests. It is expected that higher increases in
temperature will be obtained in the field and should be
carefully monitored during the field trials.
6. During its operation, the differential pressure at the
suction strainer should be checked frequently. The parallel
stand-by strainer is recommended for its continuous operation.
The twin screw multiphase pump is currently installed in
an offshore oil field in the Middle East and undergoing its
field trials. The results of these trials are the subject of a future
paper.
Acknowledgement
The authors would like to thank the Technology Research
Center of the Japan National Oil Corporation and the pump's
manufacturer Kosaka Laboratory Limited for the opportunity
to publish this work. Sincere thanks are given to Mr. K. Ikeda
and Mr. N. Hiruta, the technicians at the Kashiwazaki
Multiphase Flow Test Facility for their efforts in running the
tests and acquiring the data.