Ocean Systems Engineering, Vol. 7, No.

1 (2017) 21-38
DOI: https://doi.org/10.12989/ose.2017.7.1.021 21

Tension variations of hydro-pneumatic riser tensioner and

implications for dry-tree interface in semisubmersible
Hooi-Siang Kang1, Moo-Hyun Kim2 and Shankar S. Bhat Aramanadka3
1
Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, Johor Bahru, Johor, Malaysia
2
Department of Ocean Engineering, Texas A&M University, College Station, Texas, USA
3
Sabah Shell Petroleum Co. Ltd., Kuala Lumpur, Malaysia

(Received November 24, 2016, Revised January 25, 2017, Accepted February 5, 2017)

Abstract. In real sea environments, excessive dynamic axial tension variations can be exerted on the top-
tensioned risers (TTRs) and lead to structural integrity issues. The traditional riser-tension-variation analysis,
however, by using parametric formulation is only conditionally valid under certain strict limits and potentially
underestimates the total magnitudes of tension variations. This phenomenon is especially important for the
long stroke tensioner in dry-tree semisubmersible with larger global heave motion and longer stroke. In this
paper, the hydro-pneumatic tensioner (HPT) is modeled in detailed component-level which includes a set of
hydraulic and pneumatic components. The viscous fluid frictional effect in the HPT is considered. The main
objectives are (i) to develop a detailed tension variation model of the HPT; (ii) to identify the deviations
between the conventional parametric formulation and component-level formulation; (iii) to numerically
analyze the tension variation of long stroke tensioner in a dry-tree semisubmersible (DTS). The results
demonstrate the necessity of component-level formulation for long stroke tensioner in the development of
DTS.
Keywords: top-tension riser; hydro-pneumatic tensioner; tension variations; dry-tree semisubmersible

1. Introduction

The tensioner system of a top-tensioned riser (TTR) is designed to maintain a nearly constant
axial tension on the contact point with the TTR so that the host platform is able to move relatively
to the risers. The TTR must be tensioned so that the riser does not buckle under its own weight. The
hydro-pneumatic tensioner (HPT) systems are a form of riser tensioning mechanism used to support
TTRs on various dry-tree (well-head on deck) platforms. Conventionally, a plurality of hydraulic
cylinders with hydro-pneumatic accumulators is connected between the platform and the riser to
provide and maintain the necessary riser tension. Platform heave and horizontal motions causing
hull set-down, necessitate changes in riser length relative to the platform, which causes the
tensioning cylinders to stroke in (up-stroke) and out (down-stroke). The spring effect caused by the
fluid compression or expansion during riser stroke partially isolates the riser from the low heave
platform motions while maintaining a nearly constant riser tension. However, when the platform

Corresponding author, Professor, E-mail: m-kim3@tamu.edu

Copyright © 2017 Techno-Press, Ltd.

http://www.techno-press.com/journals/ose&subpage=7 ISSN: 2093-6702 (Print), 2093-677X (Online)
22 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

takes a significant heave motion, the compression of the fluid in the cylinders causes increased
cylinder pressure and thus increased riser tension (Crotwell and Yu 2011). The tension variations are
important in the tensioner fatigue issues (Trent 2012) and sometimes can potentially be
underestimated if only using the conventional parametric model (Gallagher et al. 2012).
In this paper, the formulation of riser tensioner is developed to identify the deviations in between
the following two approaches in the marine tensioner-riser dynamic analysis: (i) parametric
formulation (Yang 2009), and (ii) component-level formulation as the integrated component-level
effects that considers the total component effects of the HPT (Kozik 1975, Kozik and Noerager
1976, Sten et al. 2010, Gallagher et al. 2012). The effects of dynamic tension variations acting on
the TTR based on these two approaches are further simulated in a dry-tree semisubmersible (DTS)
platform to identify the effects of tension variations caused by long stroke tensioner.

2. Hydro-Pneumatic Tensioner (HPT)

There are three most common types of tensioner systems in the offshore industry, categorized as
wire-rope type, direct acting (pull-style), and ram-style (push-style), respectively. The wire-rope
type had once been popular for decades in the offshore field developments. However, due to the
complexity and limited payload ability of the wire-rope type system, the hydro-pneumatic (HP)
types have become predominant in the more recent offshore field developments. The direct-acting
(pull-style) tensioner and ram-style (push-style) tensioner can be further categorized as hydro-
pneumatic (HP) tensioner. The direct-acting (pull-style) tensioners (DAT) are mounted with the
piston rods looking down. Pressure applied to the rod side of the cylinders puts the piston rods in
tension while pulling up on the riser to generate the riser tension (Crotwell and Yu 2011). The
schematic diagram of hydro-pneumatic (HP) components is shown in Fig. 1 for a DAT system. The
tensioner acts as heave compensating system that consists of a set of hydraulic-pneumatic system
directly coupled to the riser at the tensioner ring attached to the slip joint (Sten et al. 2010). The
tensioner cylinders are symmetrically mounted under the well deck. The tensioner cylinders are
angled inboard to riser attachment points on a tensioner ring. The surface production tree or flow
control device at the top of the riser on a host floating platform can be mounted closer to the
tensioning point of the riser so that the well spacing inside the platform can be reduced. This reduces
the bending loads induced in the portion of the riser above the tension point from the dynamic
motions of the surface production equipment. Generally, the direct-acting tensioners (DATs) are
designed for relatively short strokes and therefore this system is predominately utilized in tension-
leg platform (TLP) due to the low heave characteristics of the hull, combined with the relatively
small riser length changes associated with small heave motion and set down due to the parallelogram
arrangement formed by the platform, tendons, risers, and the seafloor well pattern (Crotwell and Yu
2011).
On the other hand, the ram-style (push style) tensioners with long stroke are to be considered in
the newly developed dry-tree semisubmersible concepts (Xiang et al. 2014). The discussion of the
hydro-pneumatic system of a DAT tensioner can be generalized to the ram-style tensioner as well
since that both systems are developed from the identical hydro-pneumatic principle. The main
component is the tensioner cylinder that generates a pulling force (or pushing force in ram-style) on
the tensioner ring. The pulling force (or pushing force) is generated by a hydro-pneumatic pressure
in a closed volume. The cylinder of each tensioner is fluidly coupled, at its lower end (piston rod
side) to a hydraulic fluid reservoir pressurized by a high pressure hydro-pneumatic accumulator.
 Tension variations of hydro-pneumatic riser tensioner and … 23

Fig. 1 Schematic diagram of direct-acting tensioner (DAT) system

The top manifold is for pressurized nitrogen connected to the low pressure nitrogen pressure vessels
(NPVs) (Sten et al. 2010). On the other end, a gas (such as nitrogen or dry air) at relatively high
pressure is applied from the high pressure hydro-pneumatic accumulator to hydraulic fluid in the
cylinder rod-side chamber, driving the hydraulic fluid to push the piston upwardly in the cylinder to
retract the rod, thus pulling up the slip joint through the tensioner ring and, in turn, tensioning the
riser (Crotwell and Yu 2011). The stroke for these cylinders needs to be sufficiently long and the
volume of high pressure NPVs must be sufficiently large in order to provide nearly constant riser
tension and compensate for relative motions between the platform and the upper riser end (Sten et
al. 2010).

3. Modeling of HPT’s tension variations

The high pressure nitrogen pressure vessels (NPVs), as illustrated in Fig. 1, are designed to give
the hydraulic cylinder a desired stiffness at the required pressure setting. The pressure variations can
have a range from few bars to over 100 bars (Grønevik 2013). The relative motion between the
floating platform and the riser causes the extension and retraction of the tensioner piston rod. The
inflow of hydraulic fluid from the accumulator to the tensioner cylinder (upstroke motion) causes
retraction of the piston rod and therefore exerted pulling force on the tensioner ring. The outflow of
24 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

hydraulic fluid from the tensioner cylinder to the accumulator (down stroke motion) during the
extension of cylinder exerted large nonlinear air spring force caused by the compression of the air
volume in the accumulator and pressure vessels. Due to the large dimensions of the tensioner system,
the retracting and extending motions of tensioner cylinder involve large volumes of hydraulic-
pneumatic flow. Therefore, a large volume of compressed nitrogen in the pressure vessels is
necessary for reducing the pressure variations during this process (Sten et al. 2010).

3.1 Pressure variations in the nitrogen pressure vessel (NPV)

The high-pressure nitrogen pressure vessels (NPVs) are connected, by using pipes and hoses, to
the accumulator under a constant preload pressure controlled by the nitrogen compressor as shown
in Zone 1 in Fig. 1. The ball valves connected on these gas conduits are kept open during normal
operation to maintain small tension variations (Grønevik 2013). The total gas volume in the
boundary from the NPV to the gas domain inside the accumulator is regarded as single volume VA0
and the variation of gas volume change is represented by the volume ratio of ΔVA/VA0. The local
pressure variation in the high pressure NPV is a function of riser stroke length and that can be
calculated when the internal area and initial pressure PA0 at the equilibrium mid-position are known.
The upstroke/down-stroke motions of the tensioner rod give changes of volume in the oil/nitrogen
accumulator and NPV. The change of volume ΔVA is assumed as an adiabatic process (without the
change of heat in between the system and its environment) (Grønevik 2013). The pressure variation
ΔPA in the high pressure gas domain (from NPV to accumulator) due to the variation of gas volume
can be represented as (Kozik 1975)



PA  PA0  1  VA VA0


 1

(1)

where PA0 is the initial pressure in the high pressure NPV and γ is adiabatic gas constant, which is
1.4 of nitrogen at 15°C (Li et al. 2013). The decrement of volume -ΔVA (compression of high
pressure NPV gas domain during down-stroke motion) leads to the increment of pressure variation
ΔPA in the high pressure NPV (gas bank).

3.2 Pressure variations from NPV to accumulator

The high-pressure nitrogen pressure vessels (NPVs) are located at a certain distance away from
the accumulator on the production / drilling deck (10 m, for instance) (Sten et al. 2010) and to be
connected in gas flow connection through the pipes and valves, as illustrated in Zone 2 of Fig. 1.
The frictional losses ΔPplHA can occur as fluid (nitrogen in this case) moves through the pipes and
conduits at a flow rate qgasHA and lead to the variations in pressure applied by the tensioner system.
The pressure variations caused by the flow friction in the pipes and conduits is modeled by using
Darcy-Weisbach equation (Manring 2005). This model is derived with the assumptions that the
fluid/gas flow is fully developed along the pipe length and the effects due to the fluid/gas inertia and
fluid/gas compressibility are not taken into account, which imply that features such as water hammer
and the end effects of piping are not considered in this steady state fluid momentum conditions
(Mathworks 2009). In order to account for local resistances, such as bends, fittings, inlet and outlet
losses, and so on, all the resistances are converted into their equivalent lengths LeqHA, and then the
total length of all the resistances is added to the pipe geometrical length LHA. Therefore, the flow
 Tension variations of hydro-pneumatic riser tensioner and … 25

pressure losses ΔPplHA attributed by the pipes friction, according to Darcy-Weisbach equation, are
defined as
 gas f HA  LHA  LHA  HA HA
Ppl 
HA

eq
 qgas  qgas (2)
2
2 AHA  D
 HA 
where ρgas is the density of gas, while f HA, AHA, and DHA are the friction factor, cross sectional area,
and diameter of the pipe from the high pressure NPV (superscript H) to the accumulator (superscript
A), respectively.
Pressure losses due to friction are proportional to the flow regime-dependable friction factor f HA
and the square of the flow rate qgasHA. The friction factor f HA in Eq. (2) can be calculated based on
Moody diagram by using the Haaland equation (Haaland 1983, Mathworks 2009) as shown in Eq.
(3)




 K S Re for Re  Re L
 f  fL
f HA   fL  T  Re Re L  for Re L  Re  ReT (3)
 Re T  Re L
 2
  6.9  r  
1.11

1.8log10  Re   3.7 D    for Re  ReT
hHA

   HA   
HA
qgas  DHA
Re  (4)
AHA gas
where KS is the shape factor that characterizes the pipe cross section, Re is the Reynolds number, fL
and fT are friction factor at laminar border and turbulent border, respectively, ReL and ReT are
maximum Reynolds number at laminar flow and turbulent flow, respectively, rhHA is the height of
the internal surface roughness on the pipe from the high pressure NPV (subscript H) to the
accumulator (subscript A), and vgas is the kinematic viscosity of nitrogen gas. It is noteworthy that
the ball valves on the conduits from the NPVs to the accumulator in this paper are assumed to be
fully opened and have the same flow diameter with the pipes in connection during the normal
operation.
The preload pressure in the accumulator Ppr after considering the total effects of pressure
variation ΔPA in the high pressure NPV and pressure losses ΔPplHA due to the pipeline and valve flow
friction, as represented from the combination of Eq. (1) and Eq. (2), is stated as
Ppr  PA0  PA  PplHA (5)

 gas f HA  LHA  LHA 

 

Ppr  PA0 1  VA VA0  2 
eq
 qgas  qgas
HA HA
(6)
2 AHA  DHA 

3.3 Pressure variations from accumulator to the piston

26 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

In the hydro-pneumatic (HP) tensioner system, the accumulator consists of a pre-charged gas
domain (which is connected to the high pressure NPVs) and a fluid domain (which is connected to
the hydraulic part in the tensioner cylinder), as illustrated in Zone 3 of Fig. 1. If the hydraulic fluid
pressure at the accumulator inlet Pinlet becomes higher than the preload pressure Ppr, which is acting
on the gas-oil interface boundary, the hydraulic oil enters the accumulator chamber and compresses
the gas domain, thus storing hydro-pneumatic energy. On the other hand, a drop in the inlet fluid
pressure Pinlet during riser up-stroke motion restores the hydraulic oil back into the tensioner
cylinder. The flow rate of hydraulic oil qoil flow in / out the accumulator is calculated by using the
fluid volumetric changes induced by the motion of tensioner piston. The shut-off valve on the
conduit from the accumulator to the tensioner cylinder in this paper is assumed to be initially fully
opened and has the same passage diameter with the pipes in connection during the normal operation.
Therefore, the pressure at the piston-side chamber ppa after considering the pressure variation ΔPA
in high pressure NPV, pressure variation from the high pressure NPV to the accumulator ΔPplHA, and
pressure losses ΔPplAP due to the pipe friction and valve flow friction can be represented as
Ppa  Ppr  PplAP (7)

Ppa  PA0  PA  PplHA  PplAP (8)

   f HA  LHA  LHA 
Ppa  PA0    gas
1
 qgas  qgas
eq

HA HA


 1  V V
 
 2
2 AHA  DHA 
 A A0  (9)
oil f AP L L
AP AP

  qoil  qoil
eq
2 
2A AP  DAP 
where ρoil is the density of hydraulic oil, qoil is the flow rate of hydraulic oil, f AP, AAP, and DAP are
the friction factor, cross sectional area, and diameter of the pipe from the accumulator (superscript
A) to the tensioner piston (superscript P), respectively. The method to calculate the oil friction factor
f AP is the same as the gas friction factor f HA in the Eq. (3), by modifying the Reynold number in Eq.
(4) for the corresponding hydraulic oil inputs.
In general, the pressure loss ΔPplHA due to the friction of gas is around three orders of magnitudes
smaller than the liquid ΔPplAP, in the case of nitrogen/oil, due to the density of the fluid (Van de Ven
and Li 2009). Hence, Eq. (9) can be simplified, under a reasonable level of accuracy, into

oil f AP  L  Leq 
AP AP

 

Ppa  PA0 1  VA VA0  2   qoil  qoil (10)
2 AAP  DAP 
The industrial practices normally assume that the frictional losses of hydraulic oil in between the
accumulator and the tensioner cylinder, which is the second term of the right hand side of Eq. (10),
are to be neglected (Yang 2009). Therefore, Eq. (10) is further simplified, as the parametric
formulation, into
 
P  PA0  
1
a
(11)

 1  V V
 
p 

 A A0 
 Tension variations of hydro-pneumatic riser tensioner and … 27

The Eq. (11) is solely dependent on the pneumatic effects in the hydro-pneumatic tensioner
(HPT). The pressure at the piston-side chamber ppa is highly dependable with respect to the ratio of
volume change in the gas domain ΔVA/VA0 and the gas constant γ. The Eq. (11) can be rewritten into
the form of tension force T exerted from the tensioner cylinder piston as in the following equation
(Yang 2009, Kang 2015)

 z 
T  T0  1   (12)
 z0 

where T0 is the pretension of tensioner, z0 is the equivalent length of cylinder (which is the ratio of
total fluid volume to cylinder piston-side cross-sectional area), and Δz is the change of platform-
riser relative displacement.
The parametric formulation in Eq. (12) is reasonable if the length of connection pipes LAP is
sufficiently short (Trent 2012) or the relative velocity of the platform-riser in the heave direction is
sufficiently small, such as in the case of TLP. However, it is noteworthy that the hydraulic pressure
losses ΔPplAP through the pipes in between the accumulator and the piston in Eq. (10) must be
evaluated if the velocity of riser stroke is sufficiently large, such as under the operations of drillship,
dry-tree semisubmersible, and during the anti-recoiling and disconnection of TTR.
It is noteworthy that the formulation of Eq. (10) is based on the assumption that the fluid
compressibility and leakages inside the tensioner cylinder are assumed to be negligible. Hence, the
equation of force equilibrium on both sides of tensioner piston Ften in the hydraulic-pneumatic
cylinder can be represented as
Ften  Apa  Ppa  Abp  Ppb  FHS (13)

AP  AP 
 z  a  oil f
L  Leq
AP

Ften  Fa
HP0  1    Ap  2   qoil  qoil
 z A0  2 AAP  DAP  (14)

b  z   gas f LP  LLP  LLP 
 FHP 1   Ap   qgas  qgas  FHS
eq

b LP LP
0  2
 zB 0  2 ALP  DLP 
where FHS is the hard-stop force when the piston is approaching very closely the upper/lower limits
of the cylinder, which can be modeled as a cubic spring force (Yang 2009), zA0 and zB0 are the total
equivalent lengths of cylinder stroke in piston-side and cap-side, respectively, FHP0a and FHP0b are
the initial forces acting on the piston at the piston-side and cap-side, respectively, Apa and Apb are the
cross sectional areas of the piston at the piston-side and cap-side, respectively, ALP and DLP are the
cross sectional area and diameter of the pipe, respectively, from the tensioner piston (subscript P) to
the low pressure NPV (subscript L). Since the density of nitrogen is around three orders of
magnitudes smaller than that of hydraulic oil (Van de Ven and Li 2009), the pressure losses due to
the friction of gas can be neglected in Eq. (14) under a reasonable level of accuracy
 
AP  AP 
 z  a  oil f
L  Leq
AP
b  z 
Ften  F a
HP0 1    Ap  2   qoil  qoil  FHP0 1    FHS (15)
 z A0  2 AAP  DAP   zB 0 
It is noteworthy that the third term in the right-hand-side of Eq. (15) is normally acting as an air
28 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

spring for the up-stroke motion and, in general, FHP0b which is connected to the low-pressure NPV,
is much smaller than FHP0a.

4. Modeling of coupled dynamics of platform, riser and HPT under tension variations

The finite element (FE) model of the nonlinear HPT and the floating platform motion is
implemented in CHARM3D, a fully-coupled time-domain dynamic-analysis program for floating
bodies, mooring lines/tendons, and risers (Ran 2000, Koo et al. 2004, Yang and Kim 2011, Kang
and Kim 2012, Bae and Kim 2013, Kang et al. 2013, Kang et al. 2014, Kim and Kim 2015). The
frequency-dependent hydrodynamic coefficients and the first-order and second-order wave
excitation forces and moments are calculated by the near-field method by using WAMIT (Lee and
Newman 1991), a diffraction/radiation panel program. The corresponding forces are converted to
the time-domain terms using a two-term Volterra series expansion in CHARM3D. The frequency-
dependent radiation damping was included in the form of a convolution integral in the time domain
simulation. Viscous forces are included through the drag force term in Morison’s equation (Yang
and Kim 2010).
A detailed modeling of coupled dynamics of floating platform and riser in CHARM3D can be
found in the works of Ran (2000). In this paper, the effects of HPT’s tension variations are included.
By assuming that the low-pressure NPV is considerably lower pressurized than the high-pressure
NPVs, the air-spring reaction force FHP0b on the cap-side is significantly smaller than the pulling
force FHP0a from the piston-side, which is FHP0a>>FHP0b, then the tension Ften to be exerted by the
hydro-pneumatic tensioner (HPT) can be approximated from Eq. (15) as

 z   f AP  L  Leq 
AP AP

Ften  FHP0 1    Aa  oil   vAP  vAP (16)

 z0  2  DAP 
2
 A 
vAP  vAP  vZ vZ   a  (17)
 AAp
 
where vAP is the fluid velocity in the HPT connector pipe, and vz is the velocity of piston in the HPT
cylinder. By applying time-domain integration to Eq. (16) in the CHARM3D as derived in Kang
(2015), the coupling terms of riser finite element tangential stiffness matrix that need to be added
into the global stiffness matrix in CHARM3D after including tension variations can be represented
as
 0 0 K ijrX K ijr   0 0 K ijrX K ijr    rj   2 Ni  2  ncyn  FHPdyn 
 r          
t   0 0 0 K ij  0 0 0 0    rj  t  2 Li  t   (18)
  Xr
0
       
2   K ijXr 0 0 0   K ij 0 0 0   X j  2  2 Fi  2  2  n  FHPdyn 
   r  cyn
  K ij r  r
K ij 0 0   K ij 0 0 0     j  2 M i   
 LS TV     0 TV
where KijAB are the coupling stiffness coefficients. In CHARM3D, the symbol KijAB indicates that the
tangential stiffness coefficient for degree of freedom B in the direction A. The subscription LS refers
to the coupling connection between the floating platform and riser top node based on Eq. (12). On
 Tension variations of hydro-pneumatic riser tensioner and … 29

the other hand, the subscription TV refers to the addition of tension variations on the coupling
connection. Δt is time step of the numerical simulation, ΔXj is the change of translational motion of
the floating platform at its origin of the body coordinate system at each simulation step and Δθj is
the change of angular motion of the floating platform at each simulation step. Also, Δrj is the change
of position of the riser top node which is connected to the floating platform and Δrj’ is the change
of tangent to the riser centerline at each simulation step. Ni is the nodal resultant force on the riser
top node, Fi is the nodal resultant force acting on the floating platform, where Fi=-Ni. Moreover, Li
is nodal resultant moment on the riser top node, and Mi is the nodal resultant moment acting on the
floating platform, where Mi=Li× ri’ (Ran 2000). ncyn is the total number of hydro-pneumatic cylinder
of the riser tensioner, and FHPdyn is the dynamic part of Eq. (16), which causes the frictional effects
in the tensioner

 oil f AP  L  Leq 
AP AP

FHPdyn  Aa    v AP  v AP (19)
2  DAP 

5. Numerical analyses of HPT tension variations in regular wave

In order to identify the deviations in between the parametric formulation (Eqs. (11)-(12)) and
component-level formulation (Eq. (10) and Eq. (15)) in the analysis of TTR axial tension, the effects
of dynamic pressure/tension variations on different numerical models of hydro-pneumatic tensioner
(HPT) were investigated under the following two settings: (1) a nonlinear HPT was simulated under
a dynamic stroke and velocity for regular wave (Fig. 2); (2) the length of hydraulic pipe was varied
in order to simulate the pressure variations attributed by the frictional fluid effects (Figs. 3-5). The

Table 1 Hydraulic-pneumatic tensioner data

Pressure at LNPV* 10 bar
Volume of LNPV 4.0 m3
Pressure at HPNV* 60 bar
Volume of HNPV 9.0 m3
Length of pipe from LNPV to cylinder 30.0 m
Diameter of pipe from LNPV to cylinder 0.1 m
Length of pipe from HNPV to accumulator 30.0 m
Diameter of pipe from HNPV to accumulator 0.1 m
Length of pipe from accumulator to piston Variable
Diameter of pipe from accumulator to cylinder 0.2 m
Cylinder inner diameter 0.560 m
Piston rod diameter 0.230 m
Density of hydraulic oil 850 kg/m3
Viscosity of hydraulic oil 84.2416 cSt**
*LNPV=low pressure nitrogen pressure vessel, HNPV = high pressure nitrogen pressure vessel;
**1 cSt=1×10-6 m2/s
30 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

simulation was conducted by using the Simulink software package (Mathworks 2009) and the
CHARM3D. The dimensions and data of HPT are tabulated in Table 1.
The characteristic responses of the hydro-pneumatic tensioner (HPT) are shown in Fig. 2. The
tensioner was excited by a sinusoidal dynamic stroke Δz (from -0.50 m to 4.25 m, positive sign is
referred to down-stroke motion) and dynamic velocity (from -1.0 m/s to 1.0 m/s) in this case. The
velocity of the HPT’s piston is normally far below 1.0 m/s when the host platform is the tension-leg
platform (TLP). However, the dynamic velocity inputs in this simulation are reasonable for the cases
of drillship and dry-tree semisubmersible, in which the platform heave motion is no longer
constrained by the stiff tendons and for the case of instantaneous velocity during anti-recoiling and
TTR disconnection. The position of piston after taking the overall static loadings into account is at
around +2.0 m (positive sign is referred to downward direction in this case). If the dynamic loadings
are included, it can be seen that the piston dynamic displacement and velocity are out of phase, with
maximum velocity occurring as the piston reaches static equilibrium position (Sten et al. 2010). The
pressure in the piston-side chamber, based on Eq. (9), has the phase in between the piston
displacement and piston velocity.
The effects on the pressure variations in tensioner cylinder due to the frictional losses in the pipes
are shown in Fig. 3. The contribution of frictional losses to the pressure variations is stated by the
second term of the right hand side of Eq. (10). It can be found that the effects of friction for the 30
m length LAP long pipe is significant compared to the one with 3 m short-length LAP pipe. The
percentage of dynamic pressure variation for long pipe is ~43% while the pressure variation for the
short pipe is ~11%, with respect to mean pressure. The dynamic pressure (and tension) variations
are very sensitive to the length of hydraulic fluid pipe in the HPT. Therefore, in order to reduce
dynamic tension variations, the hydraulic pipe must be designed as short as possible, which means
that the location of the accumulator and the tensioner cylinder is recommended to be as close as
possible.
The validity and accuracy of the parametric formulation of HPT, as stated in Eq. (11) (Yang
2009, Yang and Kim 2010), was analyzed and shown in Fig. 4. The pressure variations of the 3 m
short pipe were compared with the case that the frictional losses of hydraulic oil in between the
accumulator and the tensioner cylinder ΔP pl AP are neglected. It is seen that the simplified

Fig. 2 Dynamic response of HP tensioner

 Tension variations of hydro-pneumatic riser tensioner and … 31

Fig. 3 Pressure variations due to the frictional losses in the pipes

parametric formulation is conditionally valid if the length of connection pipes LAP is sufficiently
short (3 m to 0 m), even though the dynamic velocity is large (~ -1.0 m/s to 1.0 m/s). Therefore, the
accuracy of the parametric formulation (Eq. (11)), which is a popular riser tension analysis method
in the industry, is very dependent on the physical length of the hydraulic pipe LAP of the HPT system.
The excessive dynamic tension variations are potentially to be underestimated if the HPT is built
with long hydraulic conduits and long stroke. For instance, the long-stroke tensioner is used in the
dry-tree semisubmersible.
The pressure-displacement and pressure-velocity plots with respect to the different lengths of
pipes (or hydraulic fluid traveling length) are shown in Fig. 5. It is apparent that the pressure is a
function not only of piston displacement but also of the piston velocity. Throughout the loops (in
the clock-wise direction), the pressure is greater with the extension of tensioner cylinder (positive
sign is referred to down-stroke motion in this case). The variations of hysteresis are smallest at two
end points (at ~ -0.5 m and ~ 4.25 m, respectively) and are most pronounced at the static equilibrium
point (which is at around +2.0 m in this case). It is noteworthy that the pressure variations are very
dependable to the frictional length of pipe conduits. If the length of hydraulic flow pipe is
sufficiently short (if LAP very close to 0 m), as expressed in Eq. (11), the pressure-displacement loop
is reduced to a curve line (the blue line in Fig. 5), which normally represents the HPT tension-stroke
characteristics in the industrial manufacturer manuals.

6. Numerical simulation in irregular sea condition for semisubmersible with dry-tree

interface

From the numerical analyses in the regular wave condition, it can be found that the tension
variation of HPT is dependent on the length of hydraulic fluid traveling per cycle, and the relative
velocity in between the platform and TTR. These two factors contribute minor effects in the TLP
because the tensioners in TLP are short stroke with typically in the range from 6 ft (1.83 m) to 10 ft
(3.05 m) (Leverette et al. 2013); hence the maximum allowable traveling length per cycle is small.
Moreover, the relative velocity of TLP-TTR is much smaller than one in the semisubmersible
platform. These two key factors in tension variations, however, become very significant in the
32 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

developments of dry-tree semisubmersible. The heave motion of typical dry-tree semisubmersible

(DTS) is large and long stroke tensioner is required to compensate this heave motion. Zeng et al.
(2013) proposed a deep-draft DTS comprising a two-axis symmetrical hull with draft in the range
of 100 ft (30.48 m) to 155 ft (47.24 m) to accommodate the tensioner stroke in the range of 35 ft
(10.67 m) to 45 ft (13.72 m). Poll et al. (2013) designed a Paired-Column Semisubmersible Hull
(PC-Semi) with draft of 175 ft (53.34 m) to accommodate the tensioner stroke of 28 ft (8.53 m). In
addition, Bian and Xiang (2013) also suggested a DTS hull with the draft of 150 ft (45.72 m) to
support the tensioner stroke within 35 ft (10.67 m). From these reported data of the long stroke
tensioner to be installed in the DTS, even if the accumulator is designed to be located just beside the
cylinder (Trent 2012), the hydraulic fluid is allowed to flow from 8.53 m to 13.72 m inside the length
of tensioner cylinder.
A generic type of DTS model with 12 chain-polyester-chain mooring lines, 4 steel catenary risers
(SCRs), 2 dual-barrier top tension risers (TTRs) with top tension factor 1.34 is selected for the
numerical analyses of tension variations in irregular sea condition. The principal dimensions of the

Fig. 4 Pressure variations due to the frictional pipe lengths

Fig. 5 Pressure variations in the piston-side chamber, positive stroke is referred to down-
stroke motion in this case
 Tension variations of hydro-pneumatic riser tensioner and … 33

DTS platform are tabulated in Table 2. The dimensions and data of HPT have been modified from
Table 1, and tabulated in Table 3, to customize the DTS setting where the stiffness of long-stroke
tensioner at zero-displacement is 54 kN/m. Three-hour time-simulations are carried out in a Central
Gulf of Mexico (GOM) 100-year extreme environmental condition where the metocean profiles are
listed in Table 4. Wave, wind, and current are collinear with 180° incident angle.
The time series of TTR’s stroke in the simulation of DTS under irregular sea condition is shown
in Fig. 6. The maximum down-stroke and maximum upstroke are -3.746 m and +5.713 m,
respectively. The simulated tension-stroke curve at piston side chamber of a single cylinder in the

Table 2 Principal dimensions of the generic semisubmersible model

Water depth 1219.2 m
Number of column, pontoon 4, 4
Draft 28.96 m
Total weight 29840 ST (265470 kN)
Hull group 10928 ST (97220 kN)
Deck facility 10000 ST (88964 kN)
Drilling group 3000 ST (26689 kN)
Riser pretension at the top 2216 ST (19715 kN)
Mooring load 3624 ST (32241 kN)
Single BOP tree weight 20 kips (88.96 kN)
Displacement 34810 MT (309685 kN)
*ST=short ton, MT=metric ton; data from estimation and not related to any well-existed dry-tree
semisubmersible

Table 3 Hydraulic-pneumatic tensioner data for DTS

Total number of cylinders 6
Pressure at LNPV 8.30 bar
Volume of LNPV 8.0 m3
Pressure at HNPV 49.30 bar
Volume of HNPV 2.5 m3
Length of pipe from LNPV to cylinder* 0.1 m
Diameter of pipe from LNPV to cylinder 0.1524 m
Length of pipe from HNPV to accumulator* 0.1 m
Diameter of pipe from HNPV to accumulator 0.1524 m
Length of pipe from accumulator to piston 10.0 m
Diameter of pipe from accumulator to cylinder 0.1524 m
Cross-sectional area of cylinder 0.2463 m2
Hydraulic fluid area of piston-side 0.2048 m2
Density of hydraulic oil 850 kg/m3
Viscosity of hydraulic oil 84.2416 cSt
*to be negligible in long-stroke tensioner case
34 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

Table 4 Wave, current, and wind profiles of the central-GOM 100-year condition
Significant wave height, Hs 15.8 m
Peak period, Tp 15.4 s
Overshooting parameter, γ 2.40
Direction of waves and current (collinear) 180 deg
Surface speed 2.40 m/s
Current profile Speed at mid-profile 1.80 m/s
Zero-speed depth 100.80 m
Wind 10 m elevation 48.00 m/s

Fig. 6 Stroke of TTR interface with dynamic tension variations in DTS platform

Fig. 7 Simulation of tension exerted in the piston-side chamber for single cylinder in
the long-stroke tensioner (L=10 m) for dry-tree semisubmersible case

HPT is shown in Fig. 7. The tension is ranged in [~0.60 MN to ~1.70 MN]. The maximum allowable
traveling length of the long-stroke tensioner is set to 10 m for the case of DTS under irregular sea
 Tension variations of hydro-pneumatic riser tensioner and … 35

and extreme environmental conditions. The negative sign in the DTS case is referred to the down-
stroke motion (which is different from the sign convention in the cases of regular wave simulation)
for matching the DTS global coordinate system.
As shown in Fig. 8, the tension variations exerted in single HPT cylinder at its piston side varied
from -9.0% to 11.0% from the parametric formulation results, after taking the hydraulic effects (the
second term in Eq. (15)) into account. It can be inferred from the results in Fig. 5 and Eq. (15) that
when the maximum allowable traveling length for long stroke cylinder is increased from 10 m to
13.72 m (45 ft) (Zeng et al. 2013), larger percentage of tension variations can be expected.
The tension variations in the piston-side chamber for single cylinder in the long-stroke tensioner
(L=10 m) for dry-tree semisubmersible case with respect to variations of velocity is shown in Fig.
9. As expected from the Eq. (15), larger relative velocity in between the DTS and the TTR generates
larger tension variations. The maximum velocity for DTS in this case varied from ~ -2.5 m/s to ~
2.0 m/s. From the results in Figs. 6-9, it can be further inferred that the excessive tension variations
are induced from the following sources and effects;
• The hydraulic fluid flow rate qoil which is induced by the velocity of tensioner piston, as stated
in Eq. (15). The larger tension variations caused by the larger velocity factor are reasonable for the
cases of dry-tree semisubmersible (DTS), where the platform heave motions are no longer
constrained by the stiff tendons.
• The geometrical dimension, especially the length of hydraulic pipe conduits LAP, as stated in
Eq. (15). The excessive tension variations will be exerted with longer hydraulic fluid travel between
the accumulator and tensioner cylinder, such as in the case of long-stroke tensioner.
• The ratio of gas volume ΔVA/VA0 between the displaced volume and the total gas volume stored
in the high pressure NPVs, as stated in the Eq. (10) and Eq. (11). This volume ratio must be designed
as small as possible to minimize the dynamic tension variations. Therefore, a very large volume VA0
of high pressure NPVs must be stored nearby the well bay. For the long-stroke tensioner to be
developed in the dry-tree semisubmersible platform, the required storage of high-pressurized gas
volume is even larger.

Fig. 8 Tension variations in the piston-side chamber for single cylinder in the long-stroke
tensioner (L=10 m) for dry-tree semisubmersible case with respect to variations of stroke
36 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

Fig. 9 Tension variations in the piston-side chamber for single cylinder in the long-stroke
tensioner (L=10 m) for dry-tree semisubmersible case with respect to variations of velocity

7. Conclusions

In this paper, the formulation of riser tensioner is developed to identify the deviations from the
simplified parametric formulation due to the integrated combined component-level effects. The
excessive tension variations may potentially be under-estimated when using the conventional
parametric model. The larger tension variations caused by the larger velocities are found in the case
of dry-tree semisubmersible. The dynamic tension variations are also sensitive to the hydro-
pneumatic system geometrical dimension, especially the lengths of hydraulic pipe conduits. It is
concluded that a more comprehensive formulation of long-stroke tensioner is necessary to include
the effects of excessive tension variations when the longer hydraulic fluid traveling length is
required, such as the case of long-stroke HP tensioner in the development of dry-tree
semisubmersible.

Acknowledgments

The authors would like to appreciate Sabah Shell Petroleum Co. Ltd. for the financial support in
preparing this paper.

References

Bae, Y.H. and Kim, M.H. (2013), “Influence of failed blade-pitch-control system to FOWT by aero-elastic-
control-floater-mooring coupled dynamic analysis”, Ocean Syst. Eng., 3(4), 295-307.
Bian, S. and Xiang, S. (2013), “Manage tensioner stroke for dry tree semisubmersible”, Proceedings of the
32nd International Conference on Ocean, Offshore and Arctic Engineering, Nantes, France, June.
Crotwell, G. and Yu, A. (2011), “Pull-style tensioner system for a top-tensioned riser”, U.S. Patent: 8,021,081
 Tension variations of hydro-pneumatic riser tensioner and … 37

B2, issued Sep. 20, 2011.

Gallagher, C., Williams, D. and Lang, D. (2012), “Modelling of marine riser tensioner load variations and
implications for minimum top tension settings in drilling risers”, Proceedings of the 31st International
Conference on Ocean, Offshore and Arctic Engineering, Rio de Janeiro, Brazil, June.
Grønevik, A. (2013), “Simulation of drilling riser disconnection - recoil analysis”, M.Sc. Thesis, Norwegian
University of Science and Technology, Trondheim.
Haaland, S.E. (1983), “Simple and explicit formulas for the friction factor in turbulent pipe flow”, Fluid. Eng.,
105(1), 89-90.
Kang, H.S. (2015), “Semi-active magneto-rheological damper and applications in tension leg platform / semi-
submersible”, Ph.D. Dissertation, Texas A&M University, College Station.
Kang, H.S., Kim, M.H., Bhat Aramanadka, S.S. and Kang, H.Y. (2013), “Semi-active magneto-rheological
damper to reduce the dynamic response of top-tension risers”, Proceedings of the 23rd International
Offshore and Polar Engineering Conference, Anchorage, USA, June.
Kang, H.S., Kim, M.H., Bhat Aramanadka, S.S. and Kang, H.Y. (2014), “Dynamic response control of top-
tension risers by a variable damping and stiffness system with magneto-rheological damper”, Proceedings
of the 33rd International Conference on Ocean, Offshore and Arctic Engineering, San Francisco, USA,
June.
Kang, H.Y. and Kim, M.H. (2012), “Hydrodynamic interactions and coupled dynamics between a container
ship and multiple mobile harbors”, Ocean Syst. Eng., 2(3), 217-228.
Kim, S.J. and Kim, M.H. (2015), “Dynamic behaviors of conventional SCR and lazy-wave SCR for FPSOs
in deep water”, Ocean Eng., 106, 396-414.
Koo, B.J., Kim, M.H. and Randall, R. (2004), “The effects of nonlinear multi-contact coupling with gap
between risers and guide frame on global spar motion analysis”, Ocean Eng., 31(11-12), 1469-1502.
Koos, J. (2013), “Long stroke ram-style tensioner technology”, Proceedings of the Offshore Technology
Conference, Houston, USA, May.
Kozik, T.J. (1975), “An analysis of a riser joint tensioner system”, Proceedings of the Offshore Technology
Conference, Houston, USA, May.
Kozik, T.J. and Noerager, J. (1976), “Riser tensioner force variations”, Proceedings of the Offshore
Technology Conference, Houston, USA, May.
Lee, C.-H. and Newman, J.N. (1991), “First-and second-order wave effects on a submerged spheroid”, Ship
Res., 35(3), 183-190.
Leverette, S., Bian, X. and Rijken, O. (2013), “Dry tree semisubmersibles for Gulf of Mexico”, Proceedings
of the Offshore Technology Conference, Houston, USA, May.
Li, S., Campbell, M., Howells, H. and Orsak, J. (2013), “Tension loss of hydro-pneumatic riser tensioners”,
Proceedings of the 32nd International Conference on Ocean, Offshore and Arctic Engineering, Nantes,
France, June.
Manring, N. (2005), Hydraulic Control Systems, Wiley, New Jersey, USA.
Mathworks (2009). SimHydraulics Reference 1.5, The Mathworks Inc.
Poll, P., Zou, J., Shi, S. and Kurup, N. (2013), “An evaluation of strength, fatigue and operational performance
of dry tree semisubmersible riser tensioning equipment”, Proceedings of the Offshore Technology
Conference, Houston, USA, May.
Ran, Z. (2000), “Coupled dynamic analysis of floating structures in waves and currents”, Ph.D. Dissertation,
Texas A&M University, College Station.
Sten, R., Hansen, M.R., Larsen, C.M. and Sæ vik, S. (2010), “Force variations on heave compensating system
for ultra-deepwater drilling risers”, Proceedings of the 29th International Conference on Ocean, Offshore
and Arctic Engineering, Shanghai, China, June.
Trent, D. (2012), “Tensioner system with recoil controls”, U.S. Patent: 8,157,013 B1, issued Apr. 17, 2012.
Van de Ven, J.D. and Li, P.Y. (2009), “Liquid piston gas compression”, Appl. Energy, 86(10), 2183-2191.
Xiang, S., Cao, P., Otten, J., Jiang, L. and Bian, S. (2014), “Top tensioned riser interface with dry-tree
semisubmersible”, Proceedings of the 33rd International Conference on Ocean, Offshore and Arctic
Engineering, San Francisco, USA, June.
38 Hooi-Siang Kang, Moo-Hyun Kim and Shankar S. Bhat Aramanadka

Yang, C.K. (2009), “Numerical modeling of nonlinear coupling between lines/beams with multiple floating
bodies”, Ph.D. Dissertation, Texas A&M University, College Station.
Yang, C.K. and Kim, M.H. (2010), “Linear and nonlinear approach of hydropneumatic tensioner modeling
for Spar global performance”, Offshore Mech. Artic Eng., 132(1), 011601.
Yang, C.K. and Kim, M.H. (2011), “The structural safety assessment of a tie-down system on a tension leg
platform during hurricane events”, Ocean Syst. Eng., 1(4), 263-283.
Zeng, Y., Wanvik, L-G., Wang, T., Rasmussen, S. and Lu, R. (2013), “Dry-tree semi-submersible production
and drilling unit”, U.S. Patent Application Publication: US 2013/0032076 A1, issued Feb. 7, 2013.