03 Thorp
03 Thorp
03 Thorp
by
Joseph M. Thorp
Engineering Specialist
Aramco Services Company
Houston, Texas
and
Jeffrey A. Olsen
Vice President, Software Architecture
Applied Flow Technology Corporation
Woodland Park, Colorado
ABSTRACT
Joseph M. Thorp is an Engineering
Specialist within the Technical Services A new project to increase the oil produced at Khurais, Saudi
Department of Aramco Services Company Arabia, was the focus of a study to determine the optimal pipeline
(ASC), in Houston, Texas. He has provided size and pump combination for a 90.4 mile (145.5 km) long,
technical support for Saudi Arabian Oil treated-seawater transfer line between Ain Dar and Khurais. The
Company (Saudi Aramco) projects in study evaluated the supply pumps at Ain Dar, operations with
Europe and North America, along with and without booster pumps at Khurais, and the injection pump
supporting field activities during interim configuration at Khurais. The entire system was optimally
assignments in Saudi Arabia as part of the designed to provide the required seawater flow of 2.14 million
Consulting Services Department. Mr. Thorp barrels per day (62,400 gpm) for initial operations prior to 2009
is Saudi Aramco’s designated representative to the American with the ability to be scaled up to the planned increased flow of 3.0
Petroleum Institute Subcommittee on Mechanical Equipment that million barrels per day (87,500 gpm) 10 years later.
includes Vice Chairmanship of API 610 (Centrifugal Pump) and Both first costs and life-cycle costs over 20 years were evaluated.
Chairmanship of API 682 (Seals). He is the API mechanical The tradeoff of increasing initial costs to reduce life-cycle costs was
equipment representative to the International Standards examined during the study. The pipeline material considered was
Coordinating Committee who interfaces with ISO, headquartered X-70 steel with sections welded together to form the 56 inch, 60
in Europe. Prior to joining ASC, he worked with Phillips inch, or 64 inch diameter pipe. The study assumed the pipeline
Petroleum Company. could have a maximum of two unique wall thicknesses along its
Mr. Thorp holds a B.S. degree (Mechanical Engineering) from entire length.
Michigan State University and an MBA from the University of St. The study started from a well-engineered design and through
Thomas. Mr. Thorp is a registered Professional Engineer in the optimization techniques realized a first cost savings of $37 million
State of Texas. (5 percent) and life-cycle cost savings of $104 million (12 percent).
This cost savings could have been greater except that part of the
cost reduction was used to allow the possibility for the system
Jeffrey A. Olsen joined Applied Flow to be operated at a 17 percent increased capacity without any
Technology in 1996 and as Vice President is additional expenditure for construction.
primarily responsible for the architecture INTRODUCTION AND BACKGROUND
and development of their globally marketed
software for pipe system analysis. He has The design and operating variables that most impact pipeline
contributed to software development and pump life-cycle costs have been presented in the literature (Thorp,
consulting in the areas of incompressible 2001). Users continue to rely on new and innovative approaches to
and compressible pipe flow; waterhammer satisfy consumer needs for reliable, low cost, and environmentally
analysis; pipe and pump systems friendly oil supply. Opportunities for achieving these requirements
optimization; and application deployment, have been necessitated through larger economies of scale
protection, and training. Previously, he developed computational stemming from burgeoning world oil demand. To that end, Saudi
models of mechanical systems that were used in training simulators Aramco is implementing a historically-significant project to
for the power industry. In the Thermodynamics group at General increase future production by increasing the total amount that can
Dynamics Space Systems Division, he did engineering and research be extracted from existing fields. One method to increase the
in cryogenic rocket design. amount of oil that can be extracted from an oil field is through
Mr. Olsen received a B.S. degree (Mechanical Engineering, enhanced recovery using water injection. This is accomplished by
1988) and an M.S. degree (Mechanical Engineering, 1989) from injecting high-pressure water into the reservoir thereby forcing
Brigham Young University. more oil out of the field. Many of the Saudi Arabian oil fields are
19
Table of Contents
20 PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL PUMP USERS SYMPOSIUM • 2008
located in the hot, dry, sandy desert on the eastern portion of the
country. In the desert, water is a scarce commodity; however, there
are vast amounts of water in the seas that border the country. A
large technical hurdle that must be overcome is transporting the
seawater to the oil fields over a variety of terrains using a network
of pipelines and pumping stations.
As new oil fields are developed and old ones brought back into
production, an ever growing network of seawater pipelines is pushed
farther and farther inland. This can be a costly operation, but is also a
good candidate to use optimization to find cost-cutting opportunities.
The design study described in the paper was focused on
determining the best configuration of pipeline material and sizes,
pumps, and pumping stations to meet both the current seawater
requirements and to have the flexibility to scale to accommodate
the increased demands in the future.
Optimization techniques have been used for many years on
structural applications (Schmit, 1960; Vanderplaats, 1999a) but their
use on pumping systems design is relatively new. Combining both
systemwide flow analysis and optimization techniques provides a
powerful new approach to lower costs and increase performance
during system design (Hodgson and Walters, 2002; Walters, 2002).
The existing Qurayyah seawater treatment plant, located on the
Persian Gulf, south of Ad Dammam and Dhahran, will send treated
water westward to Ain Dar located on the north portion of the
Ghawar oil field (Figure 1). The Ghawar oil field is the world’s
largest, producing about 4.5 million barrels per day, roughly 5.5
percent of the world’s daily production, and has the highest Figure 2. Details of the Ghawar and Khurais Oil Fields.
sustained oil production rate achieved by any single oil field in
world history (Croft, 2005). It stretches 174 miles from north to Meeting increasing oil demand from China, India, and other
south and 16 miles across to encompass 1.3 million acres (Durham, developing countries is the fundamental purpose of this project. To
2005). There have been 3400 wells drilled into this reservoir since accomplish this at the lowest possible life-cycle, cost numerical
the field was brought online in 1951. optimization methods were used.
SCOPE OF STUDY
General Overview
Existing operations currently use treated seawater pumped from
the Persian Gulf to Ain Dar on the northern side of the Ghawar oil
field. The idea is to use this source and pump it an additional
90.4 miles (145.5 km) westward to Khurais. The general system
guidelines called for two supply pumps at Ain Dar that would send
the seawater though a single, terrain-following pipeline. Pipe was
delivered to the site in 80 foot sections where welding of the
pipe sections and bending were performed in-situ as required by
the terrain.
When the seawater arrived at Khurais, booster pumps would
increase the pressure to meet the suction requirements of the
high-pressure injections pump. The tradeoff of incorporating
booster pumps and operating at a lower pipeline pressure versus
operating the pipeline at a higher pressure and eliminating the
booster pumps was a major design consideration.
Original Design
The study started from a well-engineered design, which will be
Figure 1. General Area Overview of Saudi Arabia Showing the referred to as the Original Case. This called for two supply pumps,
Ghawar, Ain Dar, and Khurais Oil Fields. three booster pumps and five water-injection pumps. The pipeline
was specified using X-60 steel with a 64 inch inner diameter and a
Located west of Ghawar are the Khurais, Abu Jifan and Mazalij constant wall thickness of 0.562 inches throughout. Figure 3 shows
oil fields (Figure 2). Khurais is the closest and by far the largest an overview of the original system.
being 78.9 miles (127 km) long and covering 1116 sq miles (2890
sq km) (Saudi Aramco, 2007). It is about 155 miles (250 km)
southwest of Dhahran and 186 miles (300 km) northeast of Riyadh,
the Saudi capital. Khurais is of similar structure and lies parallel to
Ghawar, but is smaller in size. Saudi Aramco has initiated a major
project aimed at increasing the production at Khurais from 300,000
to 1.2 million barrels per day by 2009 (Croft, 2005; Oil and Gas,
2006), which would make it the largest incremental increase in
world oil supply in the 75 year history of Aramco. Figure 3. Original Design with Booster Pumps.
Table of Contents
APPLICATION OF NUMERICAL OPTIMIZATION OF SEAWATER PUMPING SYSTEMS
21
The input data used in the development of the numerical The pipeline will be constructed by welding the pipe sections at the
model were based on the performance curves for two API type installation site. The Original Design called for a 64 inch diameter
BB2, 20 × 40, single-stage pumps running at 1800 rpm each pipeline with a 0.562 inch wall along the entire length
having an assumed impeller diameter of 38.07 inches (967 mm). constructed from X-60 steel, which has a yield strength of 60,000 psi.
These gas turbine driven units were each selected to provide a rated The optimized cases used X-70 steel, which has a yield strength
capacity of 31,200 gpm (7086 m3/hr). of 70,000 psi. The increased steel cost was offset by the higher
The performance and cost data for three booster pumps at strength and reduced required wall thickness. The study considered
Khurais (two operating and one spare) were entered into the model only pipeline diameters of 56, 60, and 64 inches. The smaller
and were assumed to be API type BB1, 30 × 28 single-stage, diameters required less steel and are less costly to fabricate. This
double-suction, motor-driven pumps, also having a rated capacity came at the expense of higher fluid velocities, greater pressure
of 31,200 gpm (7086 m3/hr) at 1780 rpm. drops, and increased required hydraulic horsepower.
The five injection pumps are gas turbine driven, API type BB5, Another change from the original design examined in the
20 × 20 × 21 radially split, multistage barrel pumps running at study is to allow varying wall thicknesses along the length of
4134 rpm. Each is rated at 12,500 gpm (2844 m3/hr). These the pipeline. In general the highest pressure the pipeline would
injection pumps were used in each of the cases studied. experience is at the discharge of the supply pumps—from there the
pressure tended to decrease. To meet the operating pressure
Optimized Design Configurations requirements the entire pipeline must be constructed with a wall
Various pump combinations and configurations were examined thickness dictated by the supply pump discharge pressure. To
during the study to find the one that would supply the flow minimize construction and maintenance costs a maximum of two
demands and be the most cost-efficient. Tradeoffs between wall thicknesses was allowed for the optimized designs.
operating pressures, number of pumps needed, pipeline diameter, Significant cost savings were realized by determining what wall
and pipeline wall thickness were evaluated. The system was thicknesses should be used and at what point this change should
designed to handle a continuous flowrate of 2.14 Mbbl/day take place.
(million barrels per day), or approximately 62,400 gpm, under
normal operating conditions. Figure 4 shows a typical system with Future Operations
the booster pumps eliminated. Another important consideration in the design was the ability to
increase the seawater flowrate to 3.0 Mbbl/day after 10 years. This
required the pipeline to withstand higher pressures because the
pumps needed to overcome increased frictional losses. One option
to minimize this pressure requirement was to construct a parallel
loop of pipeline for a specified length, which then recombined with
the main line at some point downstream of the supply pumps. This
decreased the flowrate in the main line and reduced the pumping
requirements due to pressure loss. Figure 5 shows a typical system
Figure 4. Optimized Design Case without Booster Pumps. using the expansion loop.
required for the higher flowrate. However, it provided the flexibility The details of the optimization methods are beyond the scope of
to vary the flow without incurring the much greater costs of this paper; however, a basic understanding of the fundamentals will
pipeline and pump modifications. help clarify the process used in designing the pipeline.
In all optimization problems there is an objective, , which
System Constraints and Parameters
is either minimized (e.g., cost) or maximized (e.g., performance).
The maximum allowable operating pressure (MAOP) can be In the case of this project the objective was to minimize the cost of
determined by Equation (1), which results in the original design the pipeline. There are several factors that make up this objective:
(60 inch, X-60 steel) having a MAOP of 758.7 psig. cost of the pipeline and coating, cost of the pumps, cost of
operation and maintenance, etc. The objective value is changed by
changing the design variables. In this case the design variables are
the diameter and thickness of the pipeline and which pumps are
being used. As these design variables are changed the model is
where: rerun to determine the new objective value or cost of the project.
P = Maximum pressure or MAOP, psig This process is repeated until the best design with the lowest cost
S = Pipe minimum yield strength, psi is found.
F = Design factor = 0.72 Now, clearly the smallest diameter and thinnest pipe would result
t = Pipe wall thickness, inches in the lowest cost. But there are several limits, or constraints, on the
D = Nominal outside pipe diameter, inches selection that the optimization must follow. These constraints
include the maximum operating pressure from Equation (2),
Based on system safety requirements, the maximum design minimum wall thickness from Equation (3), pump performance
operating pressure (MOP) cannot exceed 95 percent of the MAOP characteristics, etc. Any violation of any of the constraints results in
for the specific pipe diameter and wall thickness. Therefore, an infeasible solution and the design is discarded.
Equation (1) can be written in terms of the design pressure There are many methods of solving a constrained optimization
and wall thickness as shown in Equation (2), which was used as a problem and there is continuing research in developing even better
constraint of the optimization. and faster methods. Some of the more common ones are either
gradient-based methods or genetic algorithms methods. One of the
simplest and best known is the first-order, steepest decent method,
which is a gradient-based method. This method uses the partial
derivative of the objective function with respect to each design
Additionally, the wall thickness must satisfy a minimum project variable to determine the search direction that will result in the
design requirement shown in Equation (3). greatest change in the objective function. It then moves in that
direction to a point by changing the design variables. Finally the
objection function and partial derivatives are recalculated using the
new values of the design variables. This process of sequential steps
continues until the objective is optimized.
where: This procedure can be generalized by Equation (4), which shows
D = Nominal outside pipe diameter, inches
t = Pipe wall thickness, inches the progression of the set of design variables, , from step q to q + 1.
The optimization software tool used in this project incorporated frictional losses and to reduce maintenance, the pipeline was
several advanced methods, both gradient-based and genetic, to coated. The coating costs also varied based on diameter. The
optimize nonlinear, constrained problems as well as perform following was used as the cost of the raw steel, fabrication, and
hydraulic and cost calculations. As the optimizer engine perturbed coating:
the design variables, the hydraulic solver was run and returned the
new objective function value so the gradients could be determined. • Steel for all pipe diameters is $1.50/kg
This process was automatically repeated until the best design case • 56 inch pipe: Installation = $350/meter, Coating = $58/meter
was found. Additional details on optimization methods in general
and specific to piping networks can be found in the literature
• 60 inch pipe: Installation = $377/meter, Coating = $62/meter
(Applied Flow Technology, 2004; Hodgson and Walters, 2002; • 64 inch pipe: Installation = $450/meter, Coating = $66/meter
Vanderplaats, 1999b).
Other costs were the following:
MODEL DEVELOPMENT
• Electricity is $0.0267/kW-hr
Since the various components of the system are interdependent,
the most benefit can be obtained by optimizing the system as a • Natural gas is $0.75/MMBtu ($0.00256/kW-hr)
whole. The first step was to lay out a model of the system with • Discount rate is 8.5 percent
actual pump performance data and pipe material specifications.
The hydraulic system was modeled using a leading, commercially Configurations Cases Optimized
available network pipe and pumping system optimization tool The Original Case was modeled as a starting baseline to which
(Hodgson and Walters, 2002; Walters, 2002). This allowed for the other cases were compared. The various combinations of
rapid development and analysis of the several cases listed above. pipeline diameter and design flowrate were examined. Booster
The several pump candidates were evaluated by adding them to a pumps were required at the increased flow cases but were optional
database of pump curves and impeller diameter combinations. at the lower flowrates. Table 1 summarizes the cases that were to
Then these were swapped into the model to be evaluated. be analyzed.
The model was built to follow the general terrain of the 90.4
mile (145.5 km) pipeline, which allowed for the calculation of the Table 1. Cases Analyzed.
hydrostatic pressure at the peaks and valleys throughout the
pipeline. The optimization software tool was able to calculate the
hoop stress and give warnings when the allowable limit was
exceeded. There was an elevation gain of 889 ft (301 m) from the
supply pumps at Ain Dar (561 ft [171 m] above sea level) to the
booster/injection pumps at Khurais (1450 ft [442 m] above sea
level). The pipeline profile used is shown in Figure 6.
OPTIMIZATION RESULTS
As the matrix of cases were run it became clear that the 56 inch
pipe diameter had more pressure drop for the given flowrates than
Figure 6. Elevation Change along the Pipeline from Ain Dar to could reasonably be overcome by the supply and booster pumps.
Khurais. These cases were eliminated altogether. It was also determined that
the cost of the installation of the 64 inch pipe was large compared
The optimization tool was set to have the pipe wall thickness be to the other offsetting factors, which made these cases not viable.
the design variable and used the modern-penalty method to arrive Of all cases run the best two will be discussed here. They will be
at the best wall thickness based on the pipeline diameter. The con- referred to as Case A and Case B. Both used X-70 steel to construct
straints were automatically evaluated during the optimization a 60 inch diameter pipeline with a transition point between the two
process. Any wall thickness and pump combination that resulted in wall thicknesses. The booster pumps were eliminated in both cases
a constraint violation was rejected. Two objectives were evaluated based on the predicted NPSH at injection pumps being sufficient at
and compared: minimize first costs and minimize life-cycle costs the design flowrate.
of the overall system (materials, installation, and operation). Case A was optimized for a 2.14 Mbbl/day operating flowrate,
The cost of the raw steel was based on the weight of steel used. which resulted in a 0.60 inch wall thickness for the first 36.3 miles
This was independent of the size of the pipeline constructed. The (58.4 km) then transitioning to a 0.50 inch wall for the remaining
installation costs did vary based on the diameter. To help minimize 54.1 miles (87.1 km) of the pipeline.
Table of Contents
24 PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL PUMP USERS SYMPOSIUM • 2008
Case B was optimized for a 2.5 Mbbl/day flowrate, which resulted life including the additional looping construction. The
in a wall thickness of 0.68 inches for the first 44.8 miles (72.1 km) “Supplemental Capacity” is the additional flow beyond the 2.14
and then 0.55 inches for the remaining 45.6 miles (73.4 km). Mbbl/day design point that the system is capable of supplying
Table 2 gives a summary of the pump operating conditions without further modifications. The Original and A Cases do not
for the Original, A and B Cases at the various design flows have any built-in additional capacity, but the Case B configuration
considered. Both Cases A and B used the same API type BB3, allows the flow to increase by 0.36 Mbbl/day (17 percent) without
(20 × 20) × 24 × 28, two-stage, dual-suction supply pumps. At the additional modifications.
3.0 Mbbl/day flowrate these cases added an API type BB1, 30 × 38,
single-stage, double-suction booster pump to maintain the required Table 4. Cost Overview in $Millions.
NPSH at the injection pumps. Both the NPSH available and
required are shown for comparison in Table 2 for the booster and
injection pumps.
DISCUSSION OF RESULTS
In this study both first costs and life-cycle costs were reduced.
Often one comes at the expense of the other. In this case optimization
could reduce both from the Original Design because of several
design changes. These were the choice of steel used, reduction of
pipeline diameter, elimination of the booster pumps, and allowance
of two wall thicknesses along the pipeline. These factors are
discussed below.
Using the stronger X-70 steel for the pipeline allowed the supply
pumps to increase their discharge pressure and, therefore, the
pressure along the entire pipeline. This higher pressure allowed
Table 3 also shows the maximum pressure in the pipeline for booster pumps to be eliminated in Cases A and B at the 2.14
the section of constant wall thickness. The MAOP is based on Mbbl/day flowrate while still maintaining the required NPSH and
the pipe material and wall thickness and therefore constant in inlet pressure at the injection pumps. The increased cost of the
each section. The ratio of the maximum pressure and MAOP is X-70 steel compared to the X-60 steel used in the Original Case
also shown, which has an upper limit of 95 percent for was more than offset by the reduction in the total amount of steel
safety reasons. used and the installation costs as the pipeline diameter was reduced
Table 4 shows the optimized costs for the three cases presented. from 64 inches to 60 inches. The increased pressures in the pipeline
It is broken into two parts: the top shows the costs at the 2.14 required thicker walls, which increased the cost slightly. The total
Mbbl/day configuration and the lower part shows the additional savings on pipe installation costs was $27 million (13.5 percent)
cost required to increase the capability to 3.0 Mbbl/day. The “Total and $8 million (4 percent), respectively for Cases A and B over the
Life Cycle Cost” is the overall cost of the project over the 20-year Original Case.
Table of Contents
APPLICATION OF NUMERICAL OPTIMIZATION OF SEAWATER PUMPING SYSTEMS 25
The elimination of the booster pumps further reduced the overall At this higher flowrate the supply pumps cannot maintain a
system costs. There is a reduction in both first costs for the pumps pipeline outlet pressure high enough to meet the injection pumps
and life-cycle costs for maintenance and electricity. This saved an required NPSH, therefore Cases A and B needed booster pumps
additional $24 million over the 20-year life of the project. It is installed as well. These added an additional $15 million to the cost
important to note that the amount of energy required to push a of the system expansion. As is common in pipelines, the available
quantity of fluid down the pipeline at a certain flowrate is fairly NPSH is many, many times the NPSH required of the pumps in
constant. However, the cost of this energy varies on its source. order to ensure that the pipeline remains packed. The pump
When the electric booster pumps were eliminated, larger gas-driven selection replicates pumps that have been installed in the past.
supply pumps were required. In this situation, there is a net energy Therefore, impeller life calculations were not a major concern in
cost savings as well because the cost of electricity is about 10 times this study.
higher compared to that of natural gas. It is also important to note that part of the reduction in the cost
This analysis also allowed tradeoff studies to be made addressing of the pipe materials was the change in design requirements so that
the costs associated with building into the original design an two different wall thicknesses could be used. This allowed the pipe
increased flow capacity. The design flow point of this expanded in the downstream portion, where the pressures are not as high, to
system was 2.5 Mbbl/day. At this increased flow the overall system have a thinner wall. The minimum wall thickness is based on the
pressure drop increased and, therefore, higher supply pump nominal pipe diameter. From Equation (3), for the 60 inch pipe used
discharge pressures were required. This, in turn, required thicker in Cases A and B the minimum wall thickness is 0.5 inches. Using
pipe walls and an increase in the initial pipeline costs. With the Equation (2), this translates into a maximum pressure of 798 psig.
system optimized for this situation the overall initial cost was The transition point was determined by running the model with
increased by only 3 percent, from $682 million to $702 million, to a single wall thickness and examining the pressure profile along
achieve a 17 percent increase in system capacity. the length of the pipeline. The point at which the pressure in the
To go beyond the design flowrate to meet the expected future pipeline never exceeded this maximum pressure downstream was
demand of 3.0 Mbbl/day, all cases required additional piping to be used as the first guess at an optimal transition point. The model
installed. It was decided that the diameter of the loop would be the was then rerun using the two wall thicknesses and the pressures
same as the main line for ease of maintenance. The length of this were reexamined to verify the pressures in the thinner-walled
additional loop varied with each case. The wall thickness of the section still met the maximum pressure requirements. The transition
loop is the same as the main line because it was driven by the point was then adjusted as necessary. Finally the pump deadhead
highest pressure the lines experienced, which was at the discharge conditions were examined to ensure the wall thicknesses were also
of the supply pumps. Furthermore, the wall thickness of the main able to withstand this increased pressure. A mitigating factor was
line was based on the supply pump discharge pressure at the lower that the supply pumps were at the lowest point in the pipeline and,
flowrate. This precluded any increase in the discharge pressure the therefore, would see the greatest head pressure.
supply pumps could produce to drive the increased flow.
In a general sense, the new loop was used to reduce pressure RECOMMENDED CONFIGURATION
drop in the main line, by reducing the flowrate the main pipeline Case B was recommended and is currently under construction.
is required to carry for a certain distance. This reduction would This used a 60 inch diameter pipeline made from X-70 steel with a
balance out the increased pressure drop due to the higher wall thickness of 0.679 inches until the 44.8 miles (72.1 km) point
flowrate for the remaining length of the pipeline, which, in turn, and then a 0.551 inch wall for the remainder of the line. The
would allow the supply pumps to have a constant discharge maximum pipeline pressure for the 2.14 Mbbl/day design flow was
pressure. Since the main line and the expansion loop were both 967 psig at the supply pump discharge, which was 85 percent of the
fed by the same supply pump discharge header and had the same maximum allowable operating pressure.
geometry, they had the same pressure drop and flowrate. After Figure 7 shows the static pressure along the length of the
the two lines were merged, the main pipeline experienced the pipeline for the initial design point flowrate of 2.14 Mbbl/day. The
full, increased flowrate and, therefore, had a higher pressure drop increase in pressure between 15 and 30 miles was due to the
per length. pipeline elevation decreasing over this section. The pressure for the
An iterative approach was used to determine the required length Original Case is also shown for comparison.
of loop piping. Since the pressure drop is proportional to the
flowrate squared, an estimate of the pressure drop per length of
pipe at the higher flowrate could be made. The new loop was the
same diameter as the main line with the same wall thickness. This
balanced the system and caused the total flow to be roughly split
between the main line and the loop. Since the flow is known the
pressure drop per length in the loop could also be estimated.
Finally, the overall pressure drop from the supply pump discharge
to the booster pump inlet was known based on system and pump
requirements. Using this information an intelligent estimate of the
length for the loop piping can be made. This then had to be
adjusted based on the actual elevation of the connection point to
ensure the hydrostatic pressure did not exceed the pipe stress
limits. The new loop was added to the model of the system.
Based on the results the location was adjusted to verify all system
parameters were in the required limits.
The length and cost of the new loops needed for Cases A and B
were significantly smaller than that required for the Original Case.
This is due again to the higher operating pressures the X-70 steel
could withstand and the system configuration. The savings for the
two cases were $75 million (56 percent) and $82 million (61
percent), respectively. Case B required less loop piping cost Figure 7. Static Pressure along the Pipeline from Ain Dar to
because of the built-in additional capacity of the design. Khurais for a Flowrate of 2.14 Mbbl/day.
Table of Contents
26 PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL PUMP USERS SYMPOSIUM • 2008
The sharp pressure rise at the end of the Original Case curve Table 5 gives the supply pump deadhead (shutoff) conditions for
represents the booster pumps before the injection pump inlets. Case the pipeline at the supply pumps and at the transition point for the
B eliminated the booster pumps and had a higher pipeline pressure. three flowrates. Since the elevation at the transition point is 500 ft
As a result the inlet pressure at the injection pumps was 201 psig with (152 m) above the supply pumps, the local pressure is reduced
an NPSH available of 477 ft, well in excess of the required 133 ft. allowing a thinner wall. Again, the 3.0 Mbbl/day case has an
Careful inspection also shows the increased pressure drop per additional loop that rejoins the main pipeline before the wall
length of pipeline for Case B. This is a result of the smaller pipe transition point.
diameter and, therefore, higher pressure loss for the given flow rate.
Figure 8 shows the pressure along the pipeline for the future Table 5. Deadhead Conditions at the Supply Pumps and Wall
flowrate of 3.0 Mbbl/day. Both Case B and the Original Case use Transition Point.
booster pumps. The resulting pressure rise is shown at the end of
each curve.
Table 5 shows the local elevation and the head generated by the
supply pumps when they are deadheaded. The local head and
pressure taking into account the elevation is also shown. This is then
related to a required MAOP of the pipe wall based on Equation (2),
being 95 percent of the actual operating pressure. Finally, the
MAOP of the pipe wall is shown for the two wall thicknesses.
Figure 11 shows the sections of pipe as they are placed in the
Saudi Arabian desert in preparation for welding operations. Figure
12 depicts the apparatus used to bend the pipeline. This is
Figure 8. Static Pressure along the Pipeline from Ain Dar to necessary to allow the pipeline to follow the terrain.
Khurais for a Flowrate of 3.0 Mbbl/day.
Figures 9 and 10 show the pump data for the selected supply
pump and the injection pump.
Figure 9. Pump Performance Curve for the Supply Pumps (API Type Figure 11. Pipeline Sections Ready for Welding in Saudi Arabia.
BB3, (20 × 20) × 24 × 28, Two-Stage, Dual-Suction) at Ain Dar.
Figure 10. Pump Performance Curve for the Injection Pumps (API
Type BB5, 20 × 20 × 21 Radially Split, Multistage Barrel Pumps)
at Khurais. Figure 12. Pipeline Bending Process.
Table of Contents
APPLICATION OF NUMERICAL OPTIMIZATION OF SEAWATER PUMPING SYSTEMS 27
Applied Flow Technology, 2004, “AFT Mercury 5.5 User’s Guide,” Vanderplaats, G. N., 1999a, “Structural Design Optimization—
Woodland Park, Colorado. Status and Direction,” AIAA Journal Aircraft, 13, (1), pp. 11-20.
Croft, G., March 20, 2005, “Aramco Projects: A Closer Look,” Vanderplaats, G. N., 1999b, Numerical Optimization Techniques
Energy Bulletin.net, http://www.energybulletin.net/4785.html. for Engineering Design, Third Edition, Colorado Springs,
Colorado: Vanderplaats Research and Development, Inc.
Durham, L., January 2005, “The Elephant of All Elephants,”
Explorer, American Association of Petroleum Geologists, Walters, T, June 2002, “Cutting Costs in Pump and Pipe Sizing,”
http://www.aapg.org/explorer/2005/01jan/ghawar.cfm. Chemical Processing.