You are on page 1of 11

Proceedings of the 14th International Pipeline Conference

ASME IPC2022
September 26 – September 30, 2022, Calgary, Alberta, Canada

IPC2022-81585

PUMP STATION DESIGN II: A TALE OF TWO PUMP STATIONS

Jim Horner

AscenTech Engineering
New Westminster, British Columbia, Canada
jimhorner004@gmail.com

ABSTRACT NOMENCLATURE
IPC 2014-33740 [1], Pump Station Design, summarized the BEP Best Efficiency Point
work Enbridge had completed on the design and construction of CFD Computational Fluid Dynamics
65 new pump stations and the modification of a further 16 CSD Compact Station Design
existing stations. The $4.0 billion dollar scope was completed D Pipe length expressed in pipe diameters
over a 12-year period, with the last stations being completed in DF Design Frequency
2019. This paper documents the application of this body of work Dilbit Diluted Bitumen
on the Keystone XL Pipeline project for TC Energy. This work FEA Finite Element Analysis
included the design of 32 new pump stations. HPF High Pressure and Flow piping
The Keystone XL project experienced delays due to legal and Hz Frequency expressed in cycles/second
regulatory challenges which gave TC Energy the opportunity to NPS Nominal Pipe Size
reevaluate their pump station design. The previous design was MNF Mechanical Natural Frequency
piping centric and had a significant footprint. The station design PCD Piping Centric Design
developed for Enbridge was more equipment centric, and owing PCV Pressure Control Valve
to the compact layout, presented the potential for cost savings in ppmW Concentration in parts per million by weight
excess of $140 million for their project. This paper provides a SBC Small Bore Connections
unique perspective with which to evaluate the design TRB Transportation Research Board
philosophies employed by comparing the two station designs TRV Thermal Relief Valve
developed independently. VPF Vane Pass Frequency
While the basic design was proven, new work was completed Δm Lateral variation in mass flow
to validate the new application. This effort demonstrated that σT standard deviation of the tangential velocity
these pump stations are a unique piping subset. The operating
temperature is relatively modest, but the piping is exposed to 1. INTRODUCTION
high pressure and high flow rates (HPF). Prolonged operation This paper compares two pump station designs: one piping
with high turbulence and pressure can result in fatigue related centric and the other equipment centric. The piping centric
failures. These HPF piping systems require additional tools for approach uses typical plant piping layout strategies to minimize
their design. These analytical techniques include: piping stresses and provide adequate pipe lengths between
• Erosion studies to validate limiting velocities and pipe sizing various piping elements based on rules of thumb. This approach
will be referred to as a Piping Centric Design or PCD.
• Dynamic piping analysis to minimize potential vibration due
to beat frequencies and pressure pulsations The equipment centric approach is based on arranging the
• Finite Element Analysis (FEA) of small-bore connections to piping to allow better access to the equipment and provide a
minimize the potential for fatigue related damage. more compact layout. This approach will be referred to as a
Compact Station Design or CSD. The term compact is
• Computational Fluid Dynamics (CFD) studies to minimize significant in that the principle objective of this design was to
the impact of pump suction turbulence on pump performance. minimize the piping footprint to improve the potential for
Keywords: B-Gap, CFD, CSD, Erosion, FEA, Modal containment and leak detection. The philosophy is to make the
Analysis, Pulsation, MNF, SBC, VPF piping serve the station design rather than making the piping the
focus of the design effort.

1 © 2019 by ASME
Unit Bypass
Check Valve
Unit The scale for this
CSD Piping Plan Isolation Valve elevation section
(Discharge) has been increased
for clarity

Sump
Tank CSD Side Elevation Section

Drain System
Sump Tank
Both piping plans are
the same scale

PCD Piping Plan

Drain System

Figure 1: PCD and CSD Layout Comparison

This paper starts with a description and comparison of the 2.1 Site Development Requirements
two different designs followed by a review of the engineering The smaller footprint of the CSD led to a reduction in site
completed to validate the CSD approach. development requirements. This also meant that the containment
pond size could be reduced. While the positioning of the pond
2. COMPARISON OF DESIGNS varied depending on site specific conditions, the smaller size
A plan view of the two designs is included in Figure 1. A coupled with the smaller station footprint led to an increase in
casual inspection of the two designs shows that the CSD has a construction laydown space and a reduction in site development
significantly smaller footprint. Indeed, it can provide cost costs. These savings were obtained through a:
savings in excess of $4.4 million per station. The PCD design • 40% reduction in clearing, and rough grading;
has a flat piping arrangement with all piping at the same
• 25% reduction in piling requirements;
elevation. This means the piping is a barrier to access some of
the equipment. • 40% reduction in structural steel; and

The CSD creates access by elevating the pumps and • 50% reduction in cable and tray lengths.
providing an access platform. All the pumps, motors and sump
appurtenances are closer together and readily accessible from a 2.2 Material Requirements
common platform. The pumps and drivers are elevated by Table 2.2 shows that the amount of material used in the CSD
bringing the foundation mass aboveground. The 4,950 kW (6000 as compared to the PCD. It should be noted that the CSD used
HP) pump and motor weight 26,260 kg (29 tons), so the required predominately NPS 24 pipe while the PCD used NPS 30 pipe.
foundation mass is significant. The pipe sizing criteria will be discussed later, but the size
reduction is an integral part of the design.
The PCD piping arrangement requires a series of stiles to get
over the piping to access some of the critical components. Most Table 2.2: Station Material Requirements
of the foundation mass is located below grade which requires Flange
added excavation to form up the foundations instead being able Station Pipe (m) Elbows Tees
Pairs
to form from grade. It also creates added delays as the PCD (NPS 30) 455.7 44 10 19
excavation cannot be backfilled until the foundation has cured
CSD (NPS 24) 178.8 30 10 1
sufficiently to strip the forms.
2 © 2019 by ASME
The CSD has only 40% of the piping used in the PCD and it 2.5 Drain System
is smaller. The PCD requires 14 extra elbows and 18 extra flange Table 2.4 also shows the volumes that might be involved in a
pairs to facilitate prefabrication. The CSD arrangement does not complete station drain down. If the unit isolation valves are
include additional fittings to facilitate prefabrication but does closed, there is still a significant volume contained by the
include a single flange pair on the discharge header to provide a remainder of the aboveground station piping (by-pass and header
break to facilitate future expansion of the station piping. piping). There is 87.4 m3 (23,090 US Gal.) in the PCD piping
and 48.8 m3 (12,890 US Gal.) in the CSD piping. If the unit lines
2.3 Welding Requirements
are included, these volumes double. Even the minimum volumes
Table 2.3 shows that the PCD requires 166% more welding
are too great to be contained in the PCD 38 m3 (10,000 US Gal.)
than the CSD. This is due to the smaller pipe size and the lower
underground sump tank.
number of welded joints in the CSD.
Pumping the sump contents into the mainline concurrent with
Table 2.3: Welding Requirements a station drain down is problematic because the station would
Weld Length
have to be isolated from the mainline. The injection pump
Station Welded Joints connection is normally inside the station isolation valves on the
(cm/in)
station suction. Connecting the sump reinjection directly to the
PCD (NPS 30) 269 63,500/25,000
mainline is not typical practice due to the potential consequences
CSD (NPS 24) 188 38,100/15,000 of a leak from this secondary, small-bore mainline connection.
Table 2.5 gives an indication of the extent of the drain
The PCD design has 18 extra flange pairs to facilitate
systems (shown in red in Figure 1). The PCD drain header is
prefabrication because of the long spool lengths. The CSD
almost twice the length of the CSD header and the total drain
arrangement does not include additional flanges; the piping has
system has almost 4 times the amount of pipe as the CSD. The
enough natural breaks at unit isolation valves and unit check
PCD drain system has at least 16 more connections to the main
valves to allow prefabrication of the suction header. Two field
process piping, or primary containment, to facilitate a complete
welds are typically required on each pump riser and a couple
drain down. The sump cannot accommodate the total volume. A
more on the discharge header.
buried sump is also below the station piping; and susceptible to
overflow if too much of the piping is drained. This also means
2.4 Station Piping Volume
there are 16 additional low point drains or high potential leak
Table 2.4 shows the volumes contained by the station piping.
sources that can collect water and sediment.
The PCD’s volume is almost 2.5 times that of the CSD. The CSD
is less than half the volume of the comparable segments in the
Table 2.5: Drain System Piping
PCD piping except for the bypass piping. The lower volume in
the PCD by-pass piping is obtained by additional elbows used to Station System Component Sizes Length (m)
create an offset in the unit piping. The CSD design by-pass NPS 4 Header 98.8
volume could be reduced to half the PCD design by adding an
10 – NPS 3 Unit Line drain connections 33.6
offset with one additional elbow.
PCD 6 – NPS 3 Lateral drain connections 34.9
Table 2.4: Station Piping Volumes 10 – NPS 2 drain connections (from pump) 109.0

Piping PCD CSD Total 386.5


Segment (m3) (US Gal.) (m3) (US Gal.) NPS 4 Header 52
Unit Lines 124.2 32,810 38.5 10,150 CSD 10 - NPS 2 drain laterals (from pump) 54
By-Pass 8.8 2,310 10.6 2,800 Total 106
Headers 78.6 20,750 38.2 10,150
The entire volume of station piping could not be contained in
Total 211.6 55,860 136.1 23,100 a typical sump; therefore, auxiliary containment would be
required to purge the station piping or any significant portion.
The most significant volume is the 124.2/5 = 24.8 m3 (6,560 This would take the form of vacuum trucks and/or temporary
US Gal.) per pump contained in the PCD unit lines compared to tanks. The realization that the station piping needed to be
the 38.5/5 = 7.7 m3 (2,035 US Gal.) per pump contained in the drained very infrequently led to a rethinking of the drain system.
CSD. The PCD unit lines are flat, therefore all the 24.8 m3 would
The CSD has simplified the drain system and eliminated the
have to be purged to allow simple pump maintenance such as a
additional drain connections. The drain system is limited to the
seal replacement. Due to the elevated positioning of the pumps
collection of fluid from pump case drain downs, leakage from
in the CSD, only about 1.9 m3 (500 US Gal.) of the 7.7 m3 (2,035
the pump seals and collection of minor discharges from thermal
US Gal.) needs to be purged to drain the pump. This in turn
relief valves. This requires a smaller sump and eliminates
reduces the required sump volume.
several rarely used, high risk drain connections.

3 © 2019 by ASME
The elevated pumps allow the drain header to be kept The magnitude of the pressure pulsations is at a minimum
aboveground and supported from the pump platform. This near the pump’s BEP (best efficiency point) flow rate and
eliminates any buried station piping. The elevated pumps also increases as flow rates vary from the BEP. The pulsation level is
allow for gravity drainage to an aboveground sump tank. This affected by several pump design parameters such as:
minimizes the potential for a sump tank overflow as most of the
• The radial gap at the impeller/vane tip (B Gap, Figure 2);
piping is below the sump tank. Only volumes from the pumps
and their adjacent piping can be drained to the sump by gravity. • The impeller vane design and geometry;
3. FLUID DYNAMICS
The size of pipeline pumps has increased significantly as • The number of vanes and volute lips; and
pipeline capacities have increased. The power level of an early • The piping layout (e.g. acoustic resonances, intake swirl).
station, pumping 1,660 m3/hr (250,000 BPD) was about 6 MW
(8,000 HP). 26 MW (35,000 HP) stations are now common. As
Volute
the power levels increased, the misdirected energy resulting from Lip
pumping inefficiencies increased proportionately. This
manifests as temperature increases, noise, pressure pulsations
B-Gap
and vibration.
The power piping codes have identified a special class of
piping referred to as High Energy Piping (HEP). This includes
piping operating above 400°C (750°) and 7,100 kPa (1,025 psi).
Prolonged operation at elevated temperatures and pressures can
result in metallurgical degradation and creep related failures.
Pump station piping systems operate at modest temperatures,
but they do operate at high pressures and uniquely high flow Pump
rates. They can be categorized as high pressure and flow (HPF) Discharge
systems. Prolonged operation under high turbulence and
pressure can result in fatigue related failures. Volute
Lip
A cursory review of the two piping configurations shows that
the CSD layout has been developed to help minimize turbulence.
The CSD unit piping has only two elbows on the suction and
Figure 2: B-Gap
discharge while the PCD has three. The discharge of the CSD
pump is directed to the suction of the following pump through
two elbows instead of turning through two Tees. This improves The B-Gap has probably the most significant impact on the
the suction header flexibility while reducing the pressure drop magnitude of pressure pulsations. Kumar [2] reported a 62%
and turbulence. reduction in in vibration by increasing the B-Gap 51% in a single
volute, double suction pump. Guelich and Bolleter [3]
3.1 Impeller Design recommend a minimum B-Gap of 4%. Spence et al [4]
Pressure pulsations, as discussed here, are fluctuations in the recommend a minimum of 6% and Block [5] recommends 5%.
basic pressure being developed by a pump. These pulsations can The Keystone XL pump’s B-Gap is 10%.
sometimes cause damage, but do not receive a lot of attention
because they aren’t typically detrimental to the pump’s test-stand The vane tip design also has a significant impact on the
performance. The severity of pressure pulsations cannot be pressure pulsations. Sudo [6] and others have reported pulsation
determined merely by their presence or even magnitude, but reductions as high as 75% with staggered vanes as compared to
rather by their effect upon system components or operations. inline vane tips. Figure 3 shows a 3S impeller where the vane
tips have been split and staggered so only one half passes the
A pressure pulse is developed as each impeller vanes pass the volute lip at a time. The vane tips are also skewed so that only a
volute lip reaching a maximum value when the vane tip passes portion of the vane tip passes the volute lip at a time (3S - split,
this point. When the vane tip is adjacent to the volute lip, the staggered, skewed).
maximum amount of energy in the form of liquid in motion is
directed toward the discharge nozzle. At any other point between Corbo [7] and Stearns [8] indicate that increasing the number
vane tips, the opening between the vane and volute lip is larger of vanes can reduce pressure pulsations providing there is an odd
and, consequently, some of the liquid (and inherent energy) number. An even number with a double volute pump will
“slips by” under the volute lip and is recirculated through the increase the magnitude of the pulsations as opposed vanes pass
pump. The change from minimum to maximum transfer of each volute lip simultaneously causing constructive
energy to the discharge results in a corresponding variation in the reinforcement between the two pulses. The Keystone pump has
discharge pressure which constitutes the pressure pulsation. a 3S impeller with 7 vanes.

4 © 2019 by ASME
The pump vendor, Sulzer, proposed an acceptance criterion
that favored a balanced flow distribution for their double suction
Vane
Tip Impeller pump. The criteria included:
Shroud
• The difference in mass flow between the two sides of the
vertical centerline (lateral flow variation, Δm) should be
less than 3%.
• The standard deviation of the tangential velocity or swirl
(σT) should be less than 5% of the axial velocity.
Impeller Impeller
Eye Eye The CFD model included all piping from the discharge of one
pump to the suction of the downstream pump. The axial velocity,
tangential velocity, and lateral flow variation were modeled at
several sections downstream of the last elbow into the pump
suction. ANSI/HI 9.6.6 does allow relief from their straight pipe
recommendation if a conditioning element is used. Therefore, a
Wear piping configuration with an off-the-shelf, metering system
Ring conditioning element (Canadian Pipeline Accessories model
Central
65E) was also modeled. This preliminary study showed that the
Web
baseline flow pattern did not meet the acceptance criteria. It did
find that the conditioning element was effective in reducing the
swirl sufficiently to meet the acceptance criteria.
Figure 3: 3S Impeller Design
Further modeling was completed using a 3rd party analyst,
3.2 Station Head Loss Jacobs Technology, because the pump vendor was unable to
The pressure drop through the CSD is roughly the same as model more complex geometries. It is common for pipeline
PCD design in spite the smaller piping size. An exact companies to install a permanent strainer at each pump suction
comparison is difficult because field measurements have shown consisting of a section of grating welded into a spacer ring. The
that the pressure drop is lower than predicted by modelling. grating usually consists of 25 mm (1”) by 6.4 mm (0.25”) bearing
Standard modelling methodologies use Resistance Coefficients bars on 30 mm (1 3/16”) centers. The spacer ring is the installed
(K), Equivalent Lengths (L/D) and Flow Coefficients (CV) to between flanges. An alternate design with 75 mm (3”) deep bars
characterize fitting hydraulic resistance. Individual fitting losses arranged in a grid pattern separated by 38 mm (1.5”) of clear
are added to determine the aggregate system loss. Murakami et. space was also tested (Figure 4).
Al. [9] found that multiple fittings depending on their proximity
and relative orientation can have a lower aggregate pressure drop
than can be calculated simply as the sum of individual losses.
Zardin et. Al. [10] found that “The velocity field is strongly
affected by the relative position of two elbows: when the distance
L between the elbows is less than 5D, the two elbows strongly
influence each other, and total pressure drops are very different
to the one obtained by twice the loss on one elbow”.
The unit lines consist of a concentrated group of 7 to 9 fittings
that are within 2D to 3D of each other. Measured pressures
losses are less than 50% of the calculated result. This compact
arrangement provides the added benefit of reducing the pressure
loss through the station piping.
3.3 Computational Fluid Dynamics
The CSD has less than 2D of straight pipe adjacent to the
pump inlet proceeded by 2 perpendicular elbows. ANSI/HI
9.6.6-2016 [9] recommends a minimum of 4D to 5D of straight
pipe at the pump inlet after a long radius elbow. To alleviate any
concern that a shorter length might create an adverse flow pattern
into the pump a Computational Fluid Dynamics (CFD) study was
undertaken. The original study was completed by the pump
vendor and subsequent studies were completed by 3rd party
analysts. Figure 4: Grid Conditioning Element

5 © 2019 by ASME
The grating strainer improved the flow over the baseline case 5%
but required 3D of straight pipe to meet the acceptance criteria.

Lateral Mass Flow


Variation (Δm)
The grid conditioner performed better than the CPA conditioner, 4%
Increasing Flow Rate
but both devices exceeded the acceptance criteria. Table 3.3
3%
shows that baseline case does not meet the acceptance criteria, Variation Limit
but the grid conditioner results are well below the criteria as
2%
illustrated by the streamlines in Figures 5 & 6. The pressure drop Towards Elbow
through each device was the same, about 30 kPa (4.4 psi). 1%
0.5 0.6 0.7 0.8 0.9 1.0
Table 3.3: CFD Results Distance From Elbow Outlet (D)
Baseline Case Grid Conditioner Figure 7: Lateral Mass Flow Variation (NPS 24 Pipe)
Location
Δm σT (m/s) Δm σT (m/s) Figure 7 shows another benefit of reducing pipe size. As the
1D 4.13 % 0.496 0.41 % 0.095 flow rate and associated velocity increases the lateral flow
2D 4.92 % 0.444 0.78 % 0.011 variation improves. A reduction in velocity created by an
increase in pipe size increases the lateral flow variation. This
200 cSt, median flow rate 6.86 m/s, 5% of axial velocity is 0.341
suggests the smaller size has better turn-down characteristics.
Table 3.3 results were based on a viscosity of 200 cSt and a 3.4 Pipe Sizing
flow rate of 7,154 m3/hr (1,080,000 BPD). Δm and σT increase Pipe size has a profound impact on piping design. It affects
slightly when the viscosity decreases to 100 cSt, but the results the cost, the flexibility, and has been discussed the layout
are still within the acceptance criteria. The piping design was footprint and piping volume. A reduction in pipe size from NPS
adjusted to place a flange pair at the outlet of the upper elbow 30 to 24, can reduce pump station construction costs by millions.
turning into the pump suction. The results indicate that the grid As the piping size increases it becomes more rigid which has
conditioner would be as effective if placed at the suction nozzle. significant implications for piping flexibility. Optimizing the
pipe size has many benefits for the pump station design.
A fundamental criterion used in sizing process piping is to
define the ultimate velocity to which the piping should be
subjected, referred to as the limiting velocity. Research Report
115 [12] by the Health and Safety Executive (HSE), a UK
government agency, provides a good summary of the
considerations involved in determining a limiting velocity.
If design constraints such as pressure drop, temperature rise
or power limits do not govern, the main concern in hydrocarbon
production systems is erosion. Potential erosion mechanisms
include particulate erosion, liquid droplet erosion, erosion-
Figure 5:Baseline Streamlines corrosion, and cavitation. Report 115 [12] indicates that particles
are the most common source of erosion in hydrocarbon systems.
Sand erosion in smooth and straight pipes is normally low
and typically not the limiting factor with respect to erosion risk.
The main reason for the low erosion is the low impact angle of
erosive particles (i.e. the particle streamlines tend to be parallel
to the pipe walls). The internal surfaces of fittings that change
the flow direction or accelerate the flow will be impacted by the
0D entrained particles. These fittings are evaluated separately to
1D determine the limiting velocity they might impose.
2D
Historically API RP 14E [13] has been relied on to provide a
limiting velocity. Based on the work of Salama [14] and others,
Figure 6: Grid Conditioner Streamlines Research Report 115 [12] concludes that the recommendations
Figure 7 shows another notable trend. The Δm improves as given in API RP 14E [13] are highly conservative for onshore
the position moves closer to the elbow. This agrees with the piping systems. Under the sponsorship of Det Norske Veritas, a
findings in the PRCI Report PR-471-16206 [11] on “Suction joint industry project developed the recommended practice
Piping Effects on Pump Performance Testing”. That report DNVGL-RP-0501, “Managing Sand Production and Erosion”
found that a suction piping configuration like the CSD layout [15]. This recommended practice provides a rational basis for
with the outlet of the upper elbow placed at the pump suction determining erosion risk through calculations developed for
nozzle improved the measured pump performance. governing piping components such as elbows, tees, and reducers.

6 © 2019 by ASME
The critical parameters in the erosive flow analysis include 4. DYNAMIC ANALYSIS
flow rate, fluid properties such as viscosity and density, Flexibility analysis is performed to investigate the effect of
allowable surface thickness loss (E), design life (TL), and bending moments caused by thermally induced expansion and
entrained sediment characteristics such as erosive particle contraction, and other imposed displacements. The related stress
diameter (dP), and erosive particle concentration (EP). analysis is carried out to find the sustained (primary) stresses,
displacement (secondary) stresses, pipe support loads and
Characteristics of entrained sediments, such as the size,
equipment loading due to loads caused by the internal static
shape, mass, and hardness of solid particles, are seldom reported
pressure, deadweight, and other sustained loads. These static
in standard crude oil assays. Research Report 115 indicates that
analyses are considered mandatory for large bore piping systems.
with sand exclusion measures in place that sand particle sizes
should not exceed 100 microns. The Transportation Research Dynamic analysis is performed to investigate the effect of
Board (TRB) Special Report 311 [16] found a median particle vibration and other dynamic displacements. The related stress
size of 5.6 microns in dilbit samples tested with a maximum that analysis is carried out to determine the potential for vibration
ranged from 11 to 92 microns. Therefore, a 100-micron particle induced fatigue. These dynamic analyses are a developing field
size (dP) was assumed to be sufficiently conservative. of study with some recommended practices [17-19] but their
requirement is only starting to be implemented in the major
Sediment particle concentrations are routinely tested, but
piping codes. These dynamic analyses should be considered
these results do not distinguish between erosive particles like
mandatory for HPF piping systems.
quartz sand and non-erosive particles such as clay fines and
mineral salts. TRB Special Report 311 [16] found that silicate Energy Institute [17] guidelines provide some guidance on
particles are more abundant in dilbit and accounted for 13% to when dynamic analysis should be considered for the main piping
45% of solids, but the total level of sediments did not exceed 350 system. Figure 8 illustrates the flow induced vibration
ppmW. This suggests a maximum erosive particle concentration assessment for stiffly supported piping. The assessment
of 45% of 350, = 157.5 ppmW. quantifies the turbulent energy generated by the fluid flow in the
form of a Likelihood of Failure (LOF) index. A detailed analysis
TC Energy had found a median erosion rate of 0.056 mm/yr
is recommended for piping that exceeds the medium fatigue risk
(0.0022 in/yr) in station piping elbows. The DNVGL
threshold of 0.3.
methodology estimates an erosive particle concentration of 210
ppmW yields an erosion rate of 0.056 mm/yr based on historic
1.0
operating conditions and an erosive particle size of 100 microns.
High
The acceptable erosion rate (EL,Y) can be derived from a 0.8 Fatigue Risk
required design life (TL) and the allowable thickness loss (E) per
the following relationship. 0.6
NPS 20
LOF

Medium
𝐸𝐸 𝐸𝐸 0.4 Fatigue Risk
NPS 24
𝐸𝐸𝐿𝐿,𝑌𝑌 = 𝑜𝑜𝑜𝑜 𝑇𝑇𝐿𝐿 = 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 1
𝑇𝑇𝐿𝐿 𝐸𝐸𝐿𝐿,𝑌𝑌
0.2 NPS 30
The piping specifications for PN 100, NPS 24 pipe require a Low Fatigue Risk
material Grade 414 (X60) pipe and a 12.7 mm wall thickness. 0.0
The wall thickness required to withstand Hoop Stress at MOP is 200,000 400,000 600,000 800,000 1,000,000 1,200,000
10.16 mm. Therefore, the allowable surface thickness loss or Flow (BPD)
erosion allowance (E) is 12.7 – 10.16 = 2.55 mm. Figure 8: Turbulence Induced Vibration Risk
The erosion rates calculated for elbows and reducers are
given in Table 3.4 (dP = 100 μm, EP = 210 ppmW). These results Dynamic analysis has two areas of concern. The main
were based on a flow rate of 7,154 m3/hr (1,080,000 BPD). The concern during the early design phase is the large bore or main
design life (TL) is obtained by dividing the erosion limit (E) by process piping. As the detailed design progresses, the small-bore
the erosion rate (EL,Y). The elbow erosion rate is acceptable, but connections (SBC) to the main piping are evaluated separately
the reducer wall thickness was increased to improve design life. using a different set of analytical tools.
Table 3.4: Erosion Calculation Results Historically, dynamic analysis took the form of a modal
Erosion Rate, EL,Y Design Life, TL analysis with an extended finite element model of the main
Fittings process piping to determine the piping mechanical natural
(mm/yr) (yr)
Elbow 0.052 49.0
frequencies (MNFs). This analysis ensures that the piping has
sufficient dynamic stiffness to resist shaking forces caused by
20 x 24 Reducer 0.131 19.5 equipment such as the pumps or control valves.

7 © 2019 by ASME
The dynamic flexibility evaluation of the main piping has The vane pass frequency (VPF) is the basic frequency
grown to include a forced response analysis to calculate the generated by a centrifugal pump. A pulse is generated each time
vibration and dynamic stress due to shaking forces from pressure a vane tip passes a volute lip. If there are seven (rotating) vanes
pulsations at a range of frequencies. These pressure pulsations in the impeller and one (stationary) volute lip, each complete
are typically generated by the pumps. revolution of the impeller would generate seven pulses. If it took
one second for the impeller to complete one revolution, the vane
SBCs should always be analyzed regardless of the LOF
passing frequency would be 7 Hz.
determined for the main piping due to the inherent vulnerability
of these appurtenances. The SBC FEA is employed to determine In a perfect centrifugal pump, the VPF and its harmonics would
the MNF. It also confirms that stresses, due to relative motion of be the only discrete pulsation frequencies generated. Wnek [21]
the SBCs and the piping to which they are connected, are within states that imperfections in the pump generate additional discrete
design limits. SBCs are defined as a branched connections off frequencies. Pulsations at multiples of the rotational frequency
the main piping NPS 2 and smaller [20]. will appear due to inaccuracies in the impeller's geometric
symmetry or concentricity.
4.1 Modal Analysis
Modal analysis is the study of the dynamic properties of a In practice it is challenging to manage the geometric
system in the frequency domain. Modes are inherent properties symmetry of the impeller casting, the pump case casting, and
of a structure, and are determined by the material properties their assembly. Figure 10 shows the pressure pulsations
(mass, damping, and stiffness), and boundary conditions of the measured on discharge of the same make and model pump
structure. Each mode is defined by a MNF, modal damping, and proposed for the CSD. Despite the 3S impeller design, there are
a mode shape. The purpose of a modal analysis is to find the pressure pulsations at 2 times the operating speed (30 Hz) and
shapes and frequencies at which the structure will amplify the multiples of the nominal VPF. The 2 times component is a
effect of the input forces. common pipeline pump characteristic and significant because
pulsations at these frequencies do not dissipate as quickly as the
The goal of the modal analysis is to ensure the MNF of the
higher VPF pulsations. Studies by Timouchev and Tourret [22]
large bore piping is greater than 10 Hz. This ensures that the
show that pulsations at this frequency result from a flaw or lack
piping has enough dynamic stiffness to resist shaking forces in
of symmetry with one of the impeller vanes. The errant vane will
the 1 to 7 Hz range caused by beat frequency responses. The
interact out of sequence twice a cycle in a double volute pump.
image of the mechanical model in Figure 9 shows the first mode
with a resonant response in the crossover piping at 8.8 Hz.

Figure 9: Piping Layout Mode Shape


Figure 10: Discharge Pressure Pulsations
The red arrows show the point of maximum displacement or
antinode. The pipe support at this location was stiffened and the
mode shape changed shifting the antinode to the opposite side of The pulsation variation is less than 2% of the pump ΔP, but
the crossover and increased the MNF to 14.3 Hz. these minor fluctuations in an incompressible fluid can generate
significant alternating stresses in the piping. The pulsation
4.2 Pulsation Analysis analysis of the CSD indicated that the bypass check valve could
Pressure pulses can travel through a fluid filled pipe reflect pulses emanating from the pump. Figure 11 shows the
independent of the direction of flow. These pulses travel at the pipe segments where acoustic resonances could develop.
speed of sound in the fluid which is more than 100 times the flow Changes to the optimal CSD layout had increased the pipe length
velocity. A series of pulses is said to form a standing wave on the suction of the pump. This piping configuration had an
pattern superimposed on the normal flow. These pulsations can acoustic resonance of 66 Hz. This was too close to 2 x times the
reflect off solid objects like closed valves. If the distance operating speed particularly considering that the variable
through the fluid path between the source and the reflective frequency drive could overspeed the drive motors. Since the grid
surface equals some multiple of the pulsation wavelength this conditioning element could condition the flow with less straight
fluid path is said to be acoustically resonant, and the reflected pipe, the suction piping was collapsed, reducing the suction path
pulse will constructively reinforce or amplify the incident pulse. length, and raising its acoustic resonance.

8 © 2019 by ASME
to the TC Energy’s experience with SBC issues attributed to VPF
pulsations, as follows:
• DF of 252 Hz for SBCs within the unit isolation valves
• DF of 72 Hz for SBCs within 25 feet of a PCV
• DF of 15 Hz for SBCs for all other locations.
Suction
Discharge

Interstage

Figure 11: Potential Pathways for Acoustic Resonance

An alternate solution could have been to increase the piping


stiffness by changing the piping support design and spacing.
This becomes problematic because it can be at odds with the
piping flexibility requirements. Figure 12 shows the support
spacing recommendations from the Energy Institute guidelines
[17]. It shows that the TC Energy’s spacing guidelines are
already quite stiff, leaving little room for improvement.
20
Support Spacing (m)

15

10 Figure 13: 87.2 Hz Mode Shape for TRV

5 It was assumed that piping shell mode vibrations could be


encountered with an increase in the DF to 252 Hz. This
0 necessitated development of a more complex analytical model
100 200 300 400 500 600 700
that incorporated the vibrations of the piping to which the SBCs
Pipe Outside Diameter (mm)
are attached. This provides a more accurate model and allows
Figure 12: Energy Institute Pipe Support Spacing Guideline MNF modes caused by piping vibration to be discounted if the
SBC is moving with the piping (no relative motion). Figure 13
4.3 Small Bore Connections shows the first mode of a thermal relief valve (TRV) at 87.2 Hz.
During startup, shutdown or upset conditions SBCs can be While the vibration mode is detrimental because the TRV is
subjected to a variety of additional forces and represent a weak moving relative to the pipe, it is above the 72 Hz DF for this
point on the primary containment. Increased velocity limits location, so the bracing design is considered adequate. Bracing
result in higher flow velocities with correspondingly greater of these small-bore connections raises their natural frequencies
levels of turbulent energy. Considering this and the and minimizes relative motion with the connected piping.
consequences of an SBC failure, the need for and location of 4.4 Baseplate Design
each connection should be reviewed carefully. A more complete discussion on this topic is included in
SBC analysis includes a FEA of each typical connection to “Installation of Pipeline Pumps” [23] but it should be mentioned
estimate the MNFs. Design Frequencies (DF) are selected for that the baseplate should also be analyzed for equipment pedestal
each connection depending on the proximity to pulsation resonances. The lowest resonant frequency should be at least
sources. The MNF should exceed the DF to safeguard the SBC 20% higher than 2 times running speed.
from fatigue failure. If the SBC’s MNF is less than the DF, the Standard baseplate designs are usually wider than required to
SBC is redesigned to reduce mass, minimize the normal allow mounting of ancillary equipment. This type of design
projection, add bracing, or any combination of these strategies to increases the foundation width and moves the perimeter anchor
stiffen the connection and raise the MNF. bolts further from the pump support pedestal bases. To improve
The “Design Guideline for Small Diameter Branch the baseplate stiffness the pump should be mounted separately
Connections” [18] recommends a minimum connection Design on its own baseplate. The anchor bolts should be close to the
Frequency (DF) 20% higher than 2 times running speed for pedestal bases to improve stiffness. The foundation should be
connections within 25 ft. of the pumps and a DF greater than 15 stepped such that the motor is mounted on rails grouted directly
Hz further from the pumps. These guidelines were adjusted, due to the concrete foundation as shown in Figure 14.

9 © 2019 by ASME
[3] Guelich, J.F., Bolleter, U., “Pressure Pulsations in
Centrifugal Pumps”, ASME Journal of Vibration and Acoustics,
Pump
Support pp. 272-279, April 1992.
Pedestal Motor [4] Spence, R., Amaral-Teixeira, J., ”A CFD Parametric Study
of Geometrical Variations on Pressure Pulsations and
Performance Characteristics of a Centrifugal Pump”, Computers
& Fluids, pp. 1243-1257, June 2009.
Baseplate
[5] Block, H.P., “Petrochemical Machinery Insights”, pp. 573-
574, Butterworth Heinemann, 2017
Figure 14: Stepped Foundation
[6] Sudo, S., Komatsu, T., and Kondo, M. (1980). “Pumping
5. CONCLUSION Plant Noise Reduction: Reduction of Pressure Pulsation in Pump
Initially, the main objective of this design effort was to reduce Discharge Pipe Systems.” Hitachi Review vol. 29, No. 5, pp.
station costs through the application of a more compact station 217-222.
design. Savings in the order of $4.4 million per station were [7] Corbo, M.A., Stearns, C.F., “Practical Design Against Pump
realized, but ultimately it became apparent that this HPF piping Pressure Pulsations”, Proceedings of the Twenty-Second
system required additional analysis. Other benefits included: International Pump Users Symposium, pp. 137-177, 2005
• A significant reduction in the station footprint [8] ”Rotodynamic Pumps: for pump Piping.” ANSI/HI 9.6.6-
• Elimination of the buried sump tank and all buried piping. 2016, Approved March 23, 2016.
• Improved equipment maintainability through a reduction [9] Murakami, M.; Shimizu, Y.; Shiragami, H. Studies on fluid
in drain down volumes. flow in three-dimensional bend conduits, JSME 1969, 12,
• Elimination of 16 high risk low point drain connections 1369–1379.
for 32 station (512 connections system wide). [10] Zardin, B.; Cillo, G.; Rinaldini, C.A.; Mattarelli, E.;
• A 75% reduction in the drain piping system and the Borghi, M. Pressure Losses in Hydraulic Manifolds. Energies
associated heat tracing and insulation. 2017, 10, 310.
• A method to condition flow into the pump to allow a [11] Pipeline Research Council International, PR-471-16206
reduction in straight piping into the pump suction. Report, “Suction Piping Effect on Pump Performance Testing.”
• A reduction in pipe sizing that enhanced piping flexibility August 2017.
and improved turndown performance.
[12] Barton. “Erosion in Elbows in Hydrocarbon Production
• Better access by making the station design equipment Systems.” Research Report 115, UK Health and Safety
centric instead of piping centric. Executive. 2003. ISBN-0-7176-2743-8.
• Improved design of piping and appurtenances through [13] “Recommended Practice for Design and Installation of
various dynamic analysis techniques. Offshore Production Platform Piping Systems.” API RP 14E. 5th
edition, 1991 (revised 2000)
ACKNOWLEDGEMENTS
Ghazi Bari, Jacobs Technology, Michigan [14] Salama. “An Alternative to API 14E Erosional Velocity
Ryan Champney, Enbridge, Edmonton Limits for Sand Laden Fluids.” OTC 8898, OTC Conference,
Martin DeBusman, Formerly Jacobs, Edmonton Houston, May 4 – 7, 1998. pp 721-733, 1998.
Joe Dhaliwal, Enbridge, Edmonton
Brandon Doering, TC Energy, Calgary [15] “Managing Sand Production and Erosion.” DNVGL RP
Chris Harper, Wood VDN, Calgary 0501, Det Norske Veritas Germanischer Lloyd. August 2015.
Amir Kharazi, Jacobs Technology, Michigan [16] “Effects of Diluted Bitumen on Crude Oil Transmission
Rob Ladner, Worley ERC, Edmonton Pipelines.” Transportation Research Board Special Report 311,
Kevin Oldham, Worley ERC, Edmonton National Research Council of the National Academies. 2013.
Washington, D.C. ISBN 978-0-309-28675-6
REFERENCES
[1] Horner, Jim and DeBusman, Martin. “Pump Station Design” [17] “Guidelines for the Avoidance of Vibration Induced Fatigue
Proceedings of the 10th International Pipeline Conference, Failure in Process Pipework”. Energy Institute. Second edition,
IPC2014-33740. Calgary, Alberta, September 29 – October 3, 2008.
2014. [18] “Design Guideline for Small Diameter Branch Connections,
[2] Kumar, M.N., “Vibration Analysis of Vane Pass Frequency Release 1”. Gas Machinery Research Council, Pipeline Research
Vibrations in Single Stage, Single Volute, Between Bearings Council International, Southwest Research Institute, March
Type Pumps”, International Journal of Mechanical Engineering, 2011.
pp. 85-87, May 2017.
10 © 2019 by ASME
[19] “Reciprocating Compressors for Petroleum, Chemical, and
Gas Industry Services.” ANSI/API Standards 618-2008,
Reaffirmed August 2016.
[20] Harper, Chris. “Integrity Evaluation of Small Bore
Connections (Branch Connections).” Proceedings of the 9th
Conference of EFRC, September 10th – 12th, 2014, Vienna
[21] Wnek, T.F., “Pressure Pulsations Generated by Centrifugal
Pumps”, Warren Pumps, 2007
[22] Timoushev, S., and Toureet J., “Numerical Simulation of
BPF Pressure Pulsation Field in Centrifugal Pumps”,
Proceedings of the 19th International Pump Users Symposium,
pp. 85-105, 2002.
[23] Horner, Jim and Savage, Kevin, “Installation of Pipeline
Pumps” Proceedings of the 2nd International Pipeline
Conference, Vol 2, pp 1065-1072. Calgary, Alberta, June 7 – 11,
1998.

11 © 2019 by ASME

You might also like