Simulation of Multiphase Fluid-Hammer

Effects During Well Startup and Shut-in

Guoqing Han, China University of Petroleum; Kegang Ling, University of North Dakota; Siew Hiang Khor, SPE, Schlumberger;
He Zhang, Ryder Scott Company; and Ram Kinkar Thakur, SPE, Schlumberger

Summary On the other hand, when an upstream valve in a pipeline is

In this study, well-known commercial software that is capable of closed, the fluid downstream of the valve will attempt to continue
modeling fully transient multiphase flow in wellbore and pipeline flowing, creating a vacuum that may cause the pipe to collapse or
has been used to characterize the fluid-hammer effects of well shut- implode. This problem can be particularly acute if the pipeline is
in and startup on the coupled subsurface and surface systems. The on a downhill slope.
original work was performed by applying sensitivity analysis to In the field, wellhead and bottomhole pressures do not build up
a typical production system, including well completion, wellbore, or deplete smoothly after well shut-in or startup, respectively. The
downhole equipment (e.g., packer), and the associated surface resultant erratic pressure fluctuations are common, especially in off-
equipment (i.e., flowline, riser, and valves). This study summarizes shore operations in which packers are commonly installed. The cy-
the general course of key factors that worsen the fluid-hammer ef- clic pressure surge introduced by a sudden momentum change may
fects. Fluid hammer is also known as water hammer, a shock wave unset packers, hammer tubing, and damage the well completion,
produced by the sudden stoppage of, or reduction in, fluid flow. and may cause sand-control and other flow-assurance issues that
Field operations, such as pressure-transient analysis, facility could be very costly. Therefore, this transient scenario has drawn
maintenance, and workover, require a well shut-in process. For a significant attention from the industry in the past few decades, and
typical production system, the resultant sudden rises in pressure many independent studies have been performed in this area.
can be critical because they have a direct impact on equipment (i.e., Joukowsky (1900) presented an analytical solution to estimate
unsetting of the packer) and may cause damage to instrumentation. the pressure shock (i.e., a sudden pressure rise), which has been
This paper provides estimates of the typical ratio of transient shock well-recognized and widely applied until today. Joukowsky’s equa-
in pressure and flow rate to preconditional values, and the duration tion presents wave amplitude as a function of flow velocity, fluid
of such pressure shocks. It also proposes the best location for the density, and wave propagation. However, during the well startup
shut-in valve and the length of flowline needed to reduce the fluid- and shut-in processes, changes in wellbore pressure are instanta-
hammer effects. neous, and the conventional steady-state models are not capable
This is a pioneering approach to integrate multiphase-flow mod- of describing this transient-pressure-oscillation process. Bendiksen
eling of transient fluid-hammer effects by targeting flow-assurance et al. (1991), Han et al. (2002), and Zhang and Vairavamoorthy
issues. The software used in the study is a fully transient, commer- (2005) have hence proposed different mechanistic models that con-
cial flow-assurance simulator, and it has been used extensively for sidered not only pressure, flow rate, and temperature profile along
well-dynamics studies. The selected tool enables the integrated ap- the wellbore and flowline, but also phase redistribution in the near-
proach [i.e., from sandface (bottomhole) to wellhead and topside wellbore region.
platform, accordingly], which can be applied to surface-facility de- Coats et al. (2003), Hu et al. (2007), and Zhang et al. (2010) con-
sign and can serve as guidance in field operations to avoid hydro- ducted further studies by coupling subsurface and surface models
carbon leakages. implicitly and explicitly. Their findings have been applied for in-
tegrated field-development planning and flow-assurance studies,
Introduction especially liquid-loading and liquid-slugging problems. Ran et al.
Fluid hammer is a pressure surge or wave that occurs when a fluid (1995), Chen et al. (2007), and Jiang et al. (2008) conducted a se-
(usually a liquid, but sometimes a gas) in motion is forced to stop ries of analyses on changes of wellhead pressures for gas wells, but
or change direction suddenly (i.e., momentum change). This phe- with no modeling efforts and no proposed guidelines for field oper-
nomenon commonly occurs when a valve is closed unexpectedly at ations. Zhong et al. (2009) analyzed the impact of in-situ tempera-
the end of a pipeline system and a pressure wave propagates in the ture on pressure distribution along the wellbore through modeling
flowline. It may also be known as hydraulic shock. This pressure efforts in two cases—well shut-in and steady-state production. This
wave can cause major problems, from noise and vibration to flow- model was developed from both the mass- and energy-conservation
line rupture. If the pipeline is closed swiftly at the outlet (down- equations. Guo et al. (2010) presented a study of phase redistribu-
stream), the mass of fluid before the closure is still moving forward tion for acidic gas fluid after well shut-in, which took into consid-
with a certain velocity, building up high pressure and shock waves. eration temperature as an important factor. Ghidaoui et al. (2005)
These may cause a loud bang or repetitive banging (as the shock presented a comprehensive overview of both historic developments
waves travel back and forth), which could cause pipeline rupture. and present-day research and practice in the field concerning hy-
draulic transients. They had special discussions on mass and mo-
mentum equations for 1D and 2D flows, numerical solutions for
1D and 2D problems, wall shear-stress models, turbulence models,
Copyright © 2013 Society of Petroleum Engineers
boundary conditions, and future practical and research needs con-
This paper (SPE 160049) was accepted for presentation at the SPE Asia Pacific Oil and cerning water hammer. Chin and Xu (2001) demonstrated the use
Gas Conference and Exhibition, Perth, Australia, 22–24 October 2012, and revised for
publication. Original manuscript received for review 14 August 2012. Revised manuscript
of transient multiphase-flow analyses in the design of a deepwater
received for review 27 January 2013. Paper peer approved 3 April 2013. production system (i.e., by means of simulations of the subsea

68 Oil and Gas Facilities  •   December 2013

high-intensity pressure-protection system and emergency shut- Momentum-Conservation Equations. For gas phase,
down with support of field case studies). Wang et al. (2008) studied
the water hammer in water injectors with a field trial in which pres-
sure pulses generated from rapid shut-ins at different well depths in
(
∂ Vg ρ gν g )
= −Vg   −
2
 ∂p  1 ∂ AVg ρ gν g( )
a soft-formation case and a perforated water injector, respectively, ∂t  ∂z  A ∂z
were recorded. Tang and Ouyang (2010) and Choi and Huang
1 Sg 1 S
(2011) studied the effect of water hammer on deepwater-injection- −λ g ρg ν g ν g − λi ρ g ν r ν r i + Vg ρ g g cos α + ψ g va − FD .
well design. However, a general guide to relieve fluid-hammer ef- 2 4A 2 4A
fects upon well startup or shut-in is still incomplete. .....................................................(7)
An integrated single-flow model (with wellbore and surface
flowline) has been used in this study to investigate the water- For liquid phase at pipe wall,
hammer effects on a generic offshore production system. The
selected fully transient commercial software is a proven flow-as-
surance simulator and is also used extensively for well-dynamics
∂ (VL ρ Lν L ) (2
 ∂p  1 ∂ AVL ρ Lν L
= −VL   −
)
studies. It has the capacity to model multiphase flow in wellbore ∂t  ∂z  A ∂z
and pipeline by solving five coupled mass-conservation equations,
1 S 1 S
three momentum-conservation equations, and one energy-balance −λL ρ L ν L ν L L + λi ρ g ν r ν r i + VL ρ L g cos α
equation for a three-phase system (Bendiksen et al. 1991). 2 4A 2 4A
VL ∂V
Mass-Conservation Equations. For gas phase, −ψ g
VL + VD
(
va −ψ e vi + ψ d vD − VL d ρ L − ρ g g L sin α . )∂z
∂
(
Vg ρ g = −
1 ∂ AV g ρ gν g
) ( )
+ ψ g + G g . ............................(1) .....................................................(8)
∂t A ∂z
For liquid droplets,
For liquid phase at pipe wall,
∂ (VD ρ Lν D )
= −VD   −
2
 ∂p  1 ∂ AVD ρ Lν D ( )
∂ 1 ∂ VL
(VL ρ L ) = − ( AVL ρ Lν L ) −ψ g − ψ e + ψ d + GL . ∂t  ∂z  A ∂z
∂t A ∂z VL + VD
VD , .......(9)
.....................................................(2) +VD ρ L g cos α − ψ g va + ψ e vi − ψ d vD + FD
VL + VD
For liquid droplets, where va=vL for yg>0 (and evaporation from the liquid film),
va=vD for yg>0 (and evaporation from the liquid droplets), and
∂ 1 ∂ ( AVD ρ Lν D ) VD va=vg for yg<0 (condensation).
(VD ρ L ) = − −ψ g + ψ e − ψ d + GD .
∂t A ∂z VL + VD
Mixture Energy-Conservation Equation.
.....................................................(3) ∂   1   1   1 2 
m g  E g + v g2 + gh  + m L  E L + v L2 + gh  + mD  E D + v D + gh 
∂t   2   2   2 
For phase transfer between phases,
∂   1   1 
V  ∂ρ g  ∂p = mg vg  H g + vg2 + gh  + mL vL  H L + vL2 + gh 
 g

+
1 − Vg  ∂ρ L 
 ∂z   2   2 
   
 ρg  ∂p  T, R s ρ L  ∂p  T, R  ∂t
 s
 1 2 
+ mD vD  H D + vD + gh  + H S + U . .................................(10)

=−
1 (
∂ AVg ρ gν g )− 1 ∂ ( AVL ρ Lν L )
 2 

Aρ g ∂z Aρ L ∂z Combined with the operating and physical data described in the

next section, the simulation results demonstrated that a pressure
shock was created by a pressure difference on two sides of the shut-
1 ∂ ( AVD ρ Lν D )  1 1  in valve. In the transient shut-in period, the flow rate could increase
− +ψ g  − 
Aρ L  
∂z  ρg ρL  by as much as 10 times that of the production rate at a steady-
state condition. Setting the closing valve at the topside platform
would cause a higher peak pressure if compared with one set at
1 1 1 . .................................................(4) the wellhead or bottomhole—this finding was in agreement with
+ Gg + GL + GD
ρg ρL ρL that derived from a specific case study without considering flow-
line and riser by Choi and Huang (2011). Another observation from
For interfacial mass-transfer rate, the study was that the presence of a short flowline could help to re-
duce the pressure surge dramatically.
  ∂R  ∂p  ∂R  ∂ p ∂z  ∂R  ∂T  ∂R  ∂T ∂ z  Apart from the closing-valve location and presence of a flow-
s
ψ g =   + s  + s  + s   line/riser, it was concluded that fluid properties had significant di-
  ∂p  T ∂t  ∂p  T ∂z ∂t  ∂T  p ∂t  ∂T  p ∂z ∂t 
rect influence on water-hammer effects. For example, a high gas/
liquid-ratio flow system resulted in a lower peak shock because the
× ( mg + mL + mD ) , .................................................................(5) presence of a compressible fluid (such as gas) would reduce the
water-hammer effect significantly. Another observation made from
where the study was that the location of the shut-in valve had little or no
impact on the duration of pressure oscillations. Finally, a slow shut-
mg
Rs = . ...............................................................(6) in operation is highly recommended because it helps to reduce the
m g + mL + mD pressure surge.

December 2013  •   Oil and Gas Facilities 69

 Platform
Sea Level Start

Riser Configure/update
model
Wellhead Flowline
Seabed

Qg,o,w pbh

Wellbore Well simulator Reservoir simulator

pbh Qg,o,w
Reservoir/Completion

Fig. 1—A generic offshore production system.

No
Generic Offshore Production System If converged?
A typical offshore production scheme, as shown in Fig. 1, is used
in the study. It consists of a topside platform, riser, flowline, tubing Yes
(wellbore), and well completion. The flow system in the study
takes the following into consideration: Finish
• Three locations are available for choke installation: the plat-
form, the wellhead, and the bottom hole. Fig. 2—Explicit coupling of subsurface and surface models.
• Packers are usually installed to avoid the produced hydro-
carbon flowing through the annulus.
• A long transportation flowline might exist between the well- ervoir model, and the reservoir model calculated the flow rate of
head and the topside platform. each phase at the interface. The flow rates could be positive or neg-
ative depending on the flow directions. The setup enabled a cyclic
To describe the inlet boundary condition accurately and to jus- pressure to be present in the wellbore once the fluid flowed back
tify the advantage of using an integrated approach for simulation and forth in the near-wellbore region (Zhang et al. 2009, 2010).
of various well flow transients, Hu et al. (2007) dynamically cou- A flow model (shown in Fig. 3) with the input data presented
pled a wellbore flow model with a near-wellbore reservoir model, in Table 1 is used in this study. Taking into account that packers
and presented an iterative simulation workflow, as shown in Fig. 2. are generally installed in a typical offshore production system, the
Bottomhole pressure is represented as pbh, and Qg,o,w represents the annulus flow is not considered in this work. The side of the valve
flow rates of gas, oil, and water, respectively. During the simula- connecting to the reservoir is referred to as the “front side,” and the
tion, the wellbore model provided the pressure boundary to the res- side of the valve connecting to the separator/platform is referred
to as the “backside.” Further, the valve is placed at the platform, if
not specified.
Platform Valve
Modeling Well Startup and Shut-in
Test
In the well shut-in process, the valve front-side pressure increases
Separator suddenly until it reaches a balance with the reservoir static pressure.
Wellhead
Pipeline
Wellhead
Valve

TABLE 1—DATA SUMMARY

Fluid gravity 30°API

Gas specific gravity 0.7
re

Water cut 0%
bo
ell

Reservoir pressure 18 MPa

lve
W

Reservoir temperature 70°C

Va
ole

Well depth 1750 m

mh
l
el
W

Riser depth 285 m

tto
Bo

Flowline length 795 m

Reservoir Casing ID 0.4 m
Tubing ID 0.254 m
Fig. 3—A generic model of an offshore production system. ID = inner diameter.

70 Oil and Gas Facilities  •   December 2013

 7.0 2400
Flow-rate peak
6.5 2100
Pressure, front of valve
6.0 1800

Flow Rate, std m3/d

Pressure, MPa Pressure, back of valve
5.5 Pressure peak 1500
Flow rate
5.0 1200

4.5 900

4.0 600

3.5 300
Startup Setting pressure
3.0 0

(a) 18,000 18,015 18,030 18,045

Time, seconds
7.0 2400

6.5 2100

6.0 1800

Flow Rate, std m3/d

Pressure, MPa

5.5 1500

5.0 1200

4.5 900
Pressure, front of valve
4.0 600
Pressure, back of valve
3.5 300
Flow rate
3.0 0
17,999.8 18,000 18,000.2 18,000.4 18,000.6 18,0008.8 18,001
(b)
Time, seconds

Fig. 4—(a) Typical in-situ pressure/flow-rate trends near valve vs. time in well-start-up process, and (b) amplification of plots of in-
situ pressure/flow-rate trends near valve vs. time during the pulse.

On the backside of the valve, because the valve cuts off the supply, valve is opened and lasts less than 0.5 seconds), the front-side pres-
the pressure increases suddenly as the choke closes, and then it de- sure decreases monotonically, whereas the backside pressure in-
creases continuously as momentum removes the fluid downstream. creases monotonically until it balances the front-side pressure.
A pressure wave (depending on the lengths of flowline and riser Afterward, both the front-side and backside pressures decline si-
downstream) can be observed. This leads to a pressure difference multaneously (i.e., after a time of 18,000.2 seconds), as illustrated
on both sides of the valve. In the well-startup process, the backside in Fig. 4b.
pressure reaches a peak value close to simultaneously, which leads Identifying the Impacting Factors. Taking the same generic
to a maximum flow rate. As the front-side pressure decreases grad- model, we perform sensitivity analysis on parameters of flow rate,
ually, the pressure difference of the valve reduces gradually, and gas/oil ratio (GOR), pressure drawdown, and operation period. The
eventually stabilizes. The flow rate declines correspondingly, and results are summarized in Table 2, which shows that
it finally reaches a stable value. This process starts from a transient • The valve pressure difference reaches its peak simultaneously
condition and ends as a steady-state scenario. with well startup.
In the well-startup process, the fluid-hammer effect is not as se- • The amplitude is proportional to the pressure drawdown in a
vere as in the well-shut-in process, because the unsteady-state flow ratio range of 0.6 to 0.7.
rate acts as the kicking factor in the transient period. On the con- • Correspondingly, the flow rate reaches its peak once the valve
trary, in the well-shut-in process, the kicking factor is the pressure pressure difference is at maximum. The amplitude can be up
difference between the two sides of the valve. to 20+ times greater than the stabilized rate. As the drawdown
or GOR increases, the ratio of flow-rate peak/stabilized values
Well-Startup Process. Characteristics of Well Startup. Taking the increases.
previously developed model, after we shut in the well for 5 hours • Because the operation time of well startup is shorter, the pres-
with the valve installed at the wellhead, a stabilized system is es- sure peak and flow-rate peak are higher. However, once the op-
tablished. Next, the well starts production in a short time period of eration time is larger than a tolerance value (20 seconds in this
10 seconds. The production pressure drawdown is 2.5 MPa. The case), the influence is not dominant.
steady production rate is calculated to be 600 m3/d. The valve pres- • Because the values for flow rate, fluid GOR, operation time,
sure and flow rate vs. time, which were measured at the wellhead, and pressure drawdown are greater, the stabilization time is
are shown in Fig. 4a. longer.
When the valve is first opened, the flow rate reaches the peak
value, and the amplitude is approximately two times greater than Well-Shut-In Process. Characteristics of Well Shut-In. Unlike
that of the stable rate. During the pulse (which occurs when the the well-startup process, an afterflow effect can be observed in the

December 2013  •   Oil and Gas Facilities 71

 TABLE 2—WELL-STARTUP SENSITIVITY-ANALYSIS RESULTS

Stabilization Time
Pressure Peak/ Flow-Rate Peak/ After Startup
Drawdown Setting Flow Rate (seconds) Remarks
3 3 3
Flow rate, Q (m /d) 100 0.71 12.43 1100 GOR=0 std m /std m
Δt=10 seconds
666 0.62 2.07 1100 Δp=1.2 MPa
3 3
GOR (std m /std m ) 0 0.62 3.64 400 Δp=2.18 MPa
3
10 0.66 14.01 2500 Q=100 m /d Δt=10 seconds
Valve operation time, 1 0.72 9.42 2500 3 3
GOR=10 std m /std m
Δt (seconds)
20 0.68 8.42 2500 Δp=2.18 MPa
3
30 0.66 8.25 2500 Q=100 m /d
40 0.64 8.00 2500
60 0.64 7.75 2500
120 0.62 7.57 2500
3 3
Drawdown, Δp (MPa) 1.2 0.704 2.33 350 GOR=0 std m /std m
3
2.18 0.701 4.12 366 Q=666 m /d
12.22 0.588 22.9 450 Δt=1 second

well-shut-in process. This flow is associated with wellbore stor- When the valve is first closed, the backside pressure has a cyclic
age. When a well is first shut in, flow from the formation into the pattern of pressure surge, and the amplitude is less than that of the
wellbore bottom hole continues (even unabated) until compression front-side pressure shock. After 5 minutes (300 seconds), the front-
of the fluids in the wellbore causes the bottomhole pressure to rise. If side pressure stabilizes. For the backside pressure oscillation, the
the wellbore fluid is highly compressible and the well rate is low, the amplitude decreases gradually, and disappears completely after a
afterflow period can be long. Conversely, high-rate wells producing longer period―10 minutes (600 seconds).
little gas have negligible afterflow periods, and, consequently, could Identifying the Impacting Factors. Taking the same generic
introduce a strong hammer effect. This pressure surge transmits from model, we perform sensitivity analysis on parameters of flow rate,
both sides of the valve. For a producer, the shock is absorbed in large GOR, pressure drawdown, and operation period. The results are
part by the reservoir tank; therefore, the front-side hammer effect is summarized in Table 3, which shows that
not as cyclic as the backside effect, but it is much stronger. • After the shut-in, the front-side pressure increases quickly;
Taking the previously developed model, after the well produces however, the backside pressure increases gradually in a cyclic
for 2 hours, it establishes a stabilized production system. Next, this pattern caused by the afterflow effect. The maximum ratio of
well is shut in by the valve at the wellhead for a short period of the surge peak is approximately 1.0 to 1.5 times that of the
10 seconds. The production pressure drawdown is 2.18 MPa. The pressure drawdown.
steady production rate is calculated to be 600 m3/d. The valve pres- • The operation period of shut-in has a strong influence on
sure and flow rate vs. time are shown in Fig. 5. hammer effects. The faster the shut-in period, the worse the

TABLE 3—WELL-SHUT-IN SENSITIVITY-ANALYSIS RESULTS

Stabilization Time
Pressure Peak/ After Startup
Drawdown (seconds) Remarks
3 3 3
Flow rate, Q (m /d) 100 1.22 650 GOR=0 std m /std m
Δt=10 seconds
666 1.52 650 Δp=1.2 MPa
3 3
GOR (std m /std m ) 0 1.12 650 Δp=2.18 MPa
3
10 1.04 650 Q=100 m /d Δt=10 seconds
Valve operation time, 1 1.18 600 3 3
GOR=10 std m /std m
Δt (seconds)
20 1.02 650 Δp=2.18 MPa
3
30 1.01 650 Q=666 m /d
40 1.01 650
60 1.01 650
120 1.01 650
3 3
Drawdown, Δp (MPa) 1.2 1.44 600 GOR=0 std m /std m
3
2.18 1.27 600 Q=666 m /d
12.22 1.05 600 Δt=1 second

72 Oil and Gas Facilities  •   December 2013

 1.3
Startup at bottom hole

Pressure at Separator, MPa

Startup at wellhead
Startup at platform
1.1

0.9

0.7
18,000 18,002 18,004 18,006 18,008 18,010
Time, seconds
Fig. 5—Typical in-situ pressure trends near valve vs. time in well-shut-in process.

pressure surge will be. However, in this case, once the time the ratio of pressure peak to drawdown decreases as the draw-
is longer than 20 seconds, the pressure surge does not play down increases, the pressure peak still increases.
a role.
• A period of 10 minutes is required for production-system stabi- Modeling Well Startup and Shut-in With a Different
lization in the typical production system. Well Scheme
• Because the values for fluid GOR, operation time, and pres- Different Valve Locations. A valve can be placed at different lo-
sure drawdown are greater, the stabilization time is longer. The cations: bottom hole, wellhead, or platform. Using the previous
results are similar to those of the well-startup study, but high model, the hammer effects can be compared in Figs. 6 through 9.
flow rates lead to severe hammer effects. Note that although The study shows that

6.5 700

6.0 600
Pressure, front of valve

Flow Rate, std m3/d

5.5 500
Pressure, MPa

Pressure, back of valve

5.0 400
Flow rate
4.5 300

4.0 200

3.5 100

3.0 0
7,000 7,500 8,000 8,500 9,000
Time, seconds

Fig. 6—Pressure surge at separator upon different location of the valve (well startup).

17
Bottomhole Pressure, MPa

Startup at bottom hole

Startup at wellhead
16
Startup at platform

15

14
18,000 18,020 18,040 18,060 18,080 18,100
Time, seconds

Fig. 7—Bottomhole-pressure surge upon different location of the valve (well startup).

December 2013  •   Oil and Gas Facilities 73

 1.05

Pressure at Separator, MPa

1

0.95
Shutin at bottom hole
0.9 Shutin at wellhead
Shutin at platform

0.85

0.8
7,200 7,220 7,240
Time, seconds

Fig. 8—Pressure surge at separator upon different location of the valve (well shut-in).

17.5
Bottomhole Pressure, MPa

16.5

Shutin at bottom hole

Shutin at wellhead
15.5 Shutin at platform

14.5
7,200 7,700 8,200 8,700 9,200 9,700
Time, seconds

Fig. 9—Bottomhole-pressure surge upon different location of the valve (well shut-in).

• In the well-startup process, opening the valve at the platform Figs. 11 and 12 further illustrate that the lengths of the riser
could create a large cyclic pressure pulse. While placing the and the flowline, respectively, have a stepwise effect on atten-
valve at the well bottom hole or the wellbore could also intro- uating ­ pressure oscillations because of the increased system
duce a cyclic pressure shock, it is relatively compromising in volume. ­ Therefore, the damping effect is proportional to the
these locations. length of the riser or the flowline. Comparing Figs. 11 and 12
• In the well-shut-in process, placing the valve at the platform confirms that the presence of a flowline can help reduce pres-
creates the greatest negative impulse on the separator. On the sure shock by diverting and buffering the shock wave. The kinks
contrary, placing the valve at the well bottom hole or the well- shown in Fig. 12 demonstrate that the diverting effect dominates
head provides a tolerable pressure shock, which has less possi- for a short flowline (i.e., with a length less than 1 m), while the
bility to damage equipment. Because the afterflow effect plays buffering effect becomes more predominant for a long flowline
a role wherever the valve is placed, the bottomhole pressure al- (i.e., with a length greater than 100 m) with the same riser con-
ways builds up monotonically and reaches the balance quickly. dition. For flowline with length ranging between 1 to 100 m, the
diverting effect is the dominant factor in shock attenuation. We
Different Flowline Setups. The design of a long distance flowline believe this observation is a result of the special diverting ef-
on a seabed is a major challenge of offshore production systems. fect arising from the perpendicular angle of the fluid-flow direc-
Apart from fluid-hammer effects, other flow-assurance aspects tion and the original hammer transmission direction. As shown
have been well noticed compared with onshore completions. These in Fig. 12, the diverting effect has a steeper slope than that of
aspects include mineral (inorganic) scaling, gas hydrates/wax/as- the buffering effect. Figs. 10 and 12 suggest that even the pres-
phaltenes organic scaling, liquid loading/slugging, pipeline pig- ence of a relatively short flowline can effectively reduce water-
ging, and emulsion. hammer effect.
Simulating the previous case with the flowline deactivated and
activated helps to identify how the presence of a flowline can affect Packer Effects. Packers are installed commonly as a characteristic
the amplitude of the pressure peak and its attenuation time. Figs. 10 of an offshore production system. Taking the same model from Fig.
through 12 demonstrate that the presence of a flowline can reduce 6, we uninstall the packer. The investigation of the pressure at the
the hammer effect dramatically. This supports the concept that a same location is shown in Fig. 13.
flowline can work as a special pressure oscillation “damper” during Comparing Fig. 13 with Fig. 8, it is clear that the pressure oscil-
well startup or shut-in. The longer the flowline is, the more signifi- lation becomes attenuated in connecting with the annulus, which
cant the damping effect, as presented in Fig. 10. develops a large system volume as a buffer to absorb the oscil-

74 Oil and Gas Facilities  •   December 2013

 1.8

Pressure at Separator, MPa

No horizontal flowline
Horizontal flowline length=400 ft
1.4 Horizontal flowline length=800 ft
Horizontal flowline length=1,600 ft

0.6
18,000 18,010 18,020 18,030 18,040 18,050 18,060
Time, seconds

Fig. 10—Sensitivity study of horizontal-flowline length to impact pressure oscillation at separator (shut-in).

lations. The gas occupying the annulus is the key factor in ab- Conclusions
sorbing pressure shock resulting from sudden shut-in. It is also A dynamic, integrated wellbore/flowline (and riser) flow model
found that the setting of the valve location does not affect the pres- was built and used to study the flow-assurance aspect of fluid-
sure ­oscillations. hammer effects, which requires attention especially for offshore

1.2

1
Pressure at Separator, MPa

0.8

0.6

0.4

0.2

0
0.1 1 10 100 1000
Length of Riser, m

Fig. 11—Pressure shock variations at different riser lengths. The valve is placed at the platform (shut-in).

1.6

1.4
Pressure at Separator, MPa

1.2

0.8

0.6

0.4

0.2

0
0.1 1 10 100 1000
Length of Flowline, m
Fig. 12—Pressure shock variations at different flowline lengths. The valve is placed at the platform (shut-in).

December 2013  •   Oil and Gas Facilities 75

 1.25

Pressure at Separator, MPa 1.245 Shutin at bottom hole

Shutin at wellhead
Shutin at platform

1.24
7,200 7,220 7,240
Time, seconds
Fig. 13—Pressure surge at separator upon different location of the valve without packers installed (shut-in).

production operation. The simulation results have demonstrated H = enthalpy, J/kg

that the proposed transient integrated approach is adequate in mod- HS = enthalpy from mass sources, J/kg
eling and simulating the fluid-hammer effects introduced during mD = VD ρL, kg/m3
well startup and shut-in. The application of the proposed integrated mg = Vg ρg, kg/m3
model has also helped to conclude the following: mL = VL ρL, kg/m3
1. Pressure drawdown has a primary effect on the pressure os- p = pressure, MPa
cillation amplitude, with a range of approximately 60 to 70% pbh = bottomhole pressure, MPa
of the drawdown value. The amplitude of pressure oscillation Q = flow rate, std m3/d
exhibits little dependence on well operating time and produc- Rs = gas/oil ratio, std m3/std m3
tion rate. In the first instant of well startup, the resulting tran- S = perimeter, m
sient flow rate can reach as high as 10 times that of steady Sf = wetted perimeter, Phase f, m
state. t = time, seconds
2. During the well shut-in, the amplitude of pressure oscillation T = temperature, °C
was caused primarily by water-hammer effects, and its value U = heat transfer per unit volume, J/m3
was dependent on various factors: the higher the GOR, the v = velocity, m/s
smaller its amplitude; and the longer the valve closing time, V = volume
the smaller the wave amplitude. In the studied case, the wave VF = volumetric fractions (F=g, L, D)
oscillations resulting from water-hammer effects disappear z = length coordinate, m
after 600 seconds (i.e., 10 minutes). α = angle with gravity vector, radians
3. A shut-in/startup valve located on the platform (especially λ = friction coefficient
near the separator) will result in severe water-hammer effects. ρ = density, kg/m3
Hence, it is advisable to place the shut-in/startup valve at the ψ = mass-transfer term, kg/m3·s
bottomhole, if possible, to reduce the hammer effects.
4. In general, the hammer effects decrease with the increase of Subscripts
volume (i.e., with a longer flowline). In addition, the hammer d = droplet deposition
effects decrease slightly over the travel distance. For a small D = droplet
flow system, the water-hammer effects can be seen travelling e = droplet entrainment
back and forth clearly, but for most cases involving high vol- f = Phase f (G,L,D)
umetric flow rates, the hammer effects do not generally come g = gas
back in a long flowline. i = interfacial
5. The fluid-hammer effect is very much dependent on fluid L = liquid
compressibility rather than volumetric flow rate. A higher o = oil
fluid compressibility will lead to smaller pressure oscillation r = relative
amplitude. In a multiphaseflow system, the total fluid com- s = superficial
pressibility is subjected to the in-situ GOR. w = water
6. Finally, the volumetric effect on fluid hammering will become
insignificant for a flow system with incompressible fluid. On Acknowledgments
the other hand, a small fraction of compressible fluid (such as The authors would like to express their gratitude to Bazlee Mat-
gas) will reduce the hammer effect significantly. zain, Michael Carney, Rodney Lessard, and Fabien Houeto for their
valuable input.
Nomenclature
A = pipe cross-sectional area, m2 References
d = diameter, m Bendiksen, K.H., Malnes, D., Moe, R. et al. 1991. The Dynamic Two-Fluid
E = internal energy per unit mass, J/kg Model OLGA: Theory and Application. SPE Prod Eng 6 (2): 171–
FD = drag force, N/m3 180. SPE-19451-PA. http://dx.doi.org/10.2118/19451-PA.
g = gravitational constant, m/s2 Chen, W., Duan, Y.-G., Liu, B.-F. et al. 2007. Abnormal Well Head Pressure
G = mass source, kg/s·m3 of High Productivity Gas Wells in Luo Jiazhai (in Chinese). Journal
h = height, m of Southwest Petroleum University 29 (1): 60–63.

76 Oil and Gas Facilities  •   December 2013

Choi, S.-K. and Huang, W.S.B. 2011. Impact of Water Hammer in Deep Zhang, H., Falcone, G., Valko, P. et al. 2010. The Investigation on Counter-
Sea Water Injection Wells. Presented at the SPE Annual Technical Current Flow in the Near-Wellbore Region under Fully-Transient
Conference and Exhibition, Denver, 30 October–2 November. SPE- Condition. J. Nat. Gas Sci. Eng. 2 (2–3): 122–131.
146300-MS. http://dx.doi.org/10.2118/146300-MS. Zhang, Y. and Vairavamoorthy, K. 2005. Analysis of transient flow in pipe-
Chin, Y.D. and Xu, Z.-G. 2001. Dynamic Processes of Multiphase Flow lines with fluid–structure interaction using method of lines. Int. J.
in Subsea Flowline/Riser Systems During ESD Valve Shut Down. Numer. Methods Eng. 63 (10): 1446–1460. http://dx.doi.org/10.1002/
Presented at the SPE Latin American and Caribbean Petroleum En- nme.1306.
gineering Conference, Buenos Aires, Argentina, 25–28 March. SPE- Zhong, H., Li, Y., and Li, C. 2009. Transient numerical simulation of gas
69418-MS. http://dx.doi.org/10.2118/69418-MS. well opening and shutting-in process. Sciencepaper Online 2009-11
Coats, B.K., Fleming, G.C., Watts, J.W. et al. 2003. A Generalized Wellbore (4): 844–848.
and Surface Facility Model, Fully Coupled to a Reservoir Simulator.
Presented at the SPE Reservoir Simulation Symposium, Houston, Guoqing Han is an associate professor at China University of Petroleum,
3–5 February. SPE-79704-MS. http://dx.doi.org/10.2118/79704-MS. Beijing. His research and teaching interests include artificial-lift design,
Ghidaoui, M.S., Zhao, M., McInnis, D.A., and Ashworthy, D.H. 2005. A flow assurance, and reservoir stimulation. Han holds BSc and MSc de-
Review of Water Hammer Theory and Practice. Appl. Mech. Rev. 58 grees in process automation and production engineering, respectively,
(1): 49–76. http://dx.doi.org/10.1115/1.1828050. from China University of Petroleum, Shandong, and a PhD degree in pe-
Guo, X., Du, Z., and Fu, D. 2010. Impact of heavy constituent precipitation troleum engineering from China University of Petroleum, Beijing.
on gas component variations in a sour gas well during the shut-in pe-
riod. Natural Gas Industry 30 (3): 60–62. Kegang Ling is an assistant professor in petroleum engineering at the
Han, G., Ioannidis, M., and Dusseault, M.B. 2002. Semi-Analytical Solu- University of North Dakota. His research interests are in the area of pro-
tions for the Effect of Well Shut Down on Rock Stability. Presented duction optimization. Ling holds a BS degree in geology from the China
at the Canadian International Petroleum Conference, Calgary, 11–13 University of Petroleum and an MS degree and a PhD degree, both in
June. PETSOC-2002-050. http://dx.doi.org/10.2118/2002-050. petroleum engineering, from the University of Louisiana at Lafayette and
Hu, B., Sagen, J., Chupin, G. et al. 2007. Integrated Wellbore/Reservoir from Texas A&M University, respectively.
Dynamic Simulation. Presented at the Asia Pacific Oil and Gas Con-
ference and Exhibition, Jakarta, Indonesia, 30 October–1 November. Siew Hiang Khor is a technical consultant with Schlumberger. She has
SPE-109162-MS. http://dx.doi.org/10.2118/109162-MS. more than 18 years of research and industry experience in flow assur-
Jiang, Y.-Q., Li, C.-Y., Cao, C.-H. et al. 2008. Reason analysis on abnormal ance, production and process simulation, fluid modeling, field develop-
pressure data from high-productivity gas well testing. Natural Gas In- ment, and production optimization, and has published seven technical
dustry 28 (7): 90–92. papers. Khor holds an MEng degree (first class honors) in chemical
Joukowsky, N. 1898. Über den hydraulischen Stoss in Wasserleitungsrohren engineering and a PhD degree from Imperial College of Science, Tech-
(On the hydraulic hammer in water supply pipes), Classe Physico- nology and Medicine, London. She is a member of SPE and the Amer-
Mathématique Volume IX, No. 5, Series VIII. St.-­Pétersbourg, Russia: ican Institute of Chemical Engineers, and has served as a program
Mémoires de l’Académie Impériale des Sciences de St.-Pétersbourg, committee member for SPE applied technology workshops.
l’Académie Impériale des Sciences de St.-Pétersbourg.
Ran, X.-Q., Chen, Q.-L., and Zhao, B.-R. 1995. Treatment method research He Zhang, a reservoir engineer at Ryder Scott Petroleum Consultants,
on wellhead pressure build up curve abnormality of gas well. Natural evaluates oil and gas reserves. He has more than 10 years of research
Gas Industry 15 (6): 24–27. and industry experience in reservoir and production simulation, pore/
Tang, Y. and Ouyang, L.-B. 2010. A Dynamic Simulation Study of Water volume/temperature characterization, and petroleum economics.
Hammer for Offshore Injection Wells to Provide Operation Guide- Zhang has published approximately 20 technical papers, and serves as
lines. Presented at the International Oil and Gas Conference and Ex- an associate editor and peer reviewer for several journals. He holds BS
hibition in China, Beijing, 8–10 June. SPE-131594-MS. http://dx.doi. degrees in both chemistry and computer application from the University
org/10.2118/131594-MS. of Science and Technology of China, and a PhD degree in petroleum en-
Wang, X., Hovem, K.A., Moos, D. et al. 2008. Water Hammer Effects on gineering from Texas A&M University.
Water Injection Well Performance and Longevity. Presented at the SPE
International Symposium and Exhibition on Formation Damage Con- Ram Kinkar Thakur is a senior petroleum engineer with Schlumberger,
trol, Lafayette, Louisiana, USA, 13–15 February. SPE-112282-MS. and has more than 17 years of experience in the oil and gas industry.
http://dx.doi.org/10.2118/112282-MS. He has worked on numerous projects, including but not limited to, field-
Zhang, H., Falcone, G., Valko, P.P. et al. 2009. Numerical Modeling of Fully- development planning, production operations and optimization, well
Transient Flow in the Near-Wellbore Region During Liquid Loading in completions and workover, facilities, flow assurance, and production
Gas Wells. Presented at the Latin American and Caribbean Petroleum simulation. Thakur holds a BS degree in petroleum engineering from
Engineering Conference, Cartagena de Indias, Colombia, 31 May–3 the Indian School of Mines, Dhanbad, India. He is a member of SPE and
June. SPE-122785-MS. http://dx.doi.org/10.2118/122785-MS. the American Institute of Chemical Engineers.

December 2013  •   Oil and Gas Facilities 77