You are on page 1of 15

Proceedings of the ASME 2009 28th International Conference on Ocean, Offshore and Arctic Engineering

Proceedings of the ASME 28th International Conference on Ocean, Offshore and Arctic Engineering
OMAE2009
OMAE2009
May 31 - June 5, 2009, Honolulu, Hawaii, USA
May 31 - June 5, 2009, Honolulu, Hawaii

OMAE2009-79759
OMAE2009-79759

STRAIGHT TALK ABOUT RISER TENSION AND MORE


Roger Chang
Engineering, Research & Computing
Houston, Texas USA

column.
ABSTRACT • Sparks: Effective tension is simply the real total force in the composite
column.
One of most confusing issues in riser engineering is the riser
• Seyed, Patel: Mathematical restatement of Archimedes principle for a
tension. The infamous effective tension equation relates it to the so-
continuous pipe.
called material tension with external and internal pressures. • Major oil company researcher: Buoyancy is a phantom force.
Controversy remains after numerous papers published trying to clarify • Major oil company researcher: Riser will not carry mud weight.
the subject, because different interpretations were presented by
different authors. Instead of explaining this ‘abstract’ equation Two most authoritative regulatory codes in riser engineering are
mathematically using the free body diagram and differential equation API RP 2RD [5] and DNV-OS-F201 [6]. For the axial stress
as done in the literatures, this paper presents a down-to-earth calculation, Section 5 Equation 8 of API RP 2RD calls for the material
interpretation that follows the riser loading history which starts with tension, but DNV specifically states that the effective tension shall be
the Effective Weight to re-derive the same equation. Four keys to used. The question then is which is correct. Sparks [3] adapted the
solve the riser tension mystery are identified; they are the hydrostatic effective stress concept from Soil Mechanics and demonstrated the
head pressure vs. applied pressure, pressure generates the pressure end same von Mises stress may be calculated by either tension. This paper
cap load vs. none generated, the vertical (top-tensioned) riser vs. bent will show that the effective stress is valid up to the riser
(catenary) riser, and the single string riser vs. multiple strings riser. pressurization; additional loads must be included in the stress
Based on these four keys, this paper will address the difference calculation afterwards.
between the effective tension and material tension and identify which
In order to fully understand the buoyancy, hence riser tension,
tension is to be used in the stress calculation. Also presented in the
author has conducted a riser tension verification study using Abaqus
paper is the driver-reactor theory developed to explain the tension load
[7]. The study included three riser configurations: vertical straight
distribution among riser strings due to Poisson’s effect with the
pipe, tilted straight pipe and curved pipe. Both open-end pipe and
applied pressure.
close-end pipe were investigated in conjunction with the above three
riser configurations. All the cases were studied using three different
INTRODUCTION types of elements: beam (pipe), shell and solid. But, only the constant
Riser tension has been one of the most confusing issues in riser pressure was studied. In preparing this manuscript, additional
engineering. For any young engineer getting into the riser design verification works were done using catenary riser for a bent pipe
and/or analysis arena, he or she will be taught that there is an subject to depth varying pressures. The ‘bold’ statements to be made
‘effective’ tension for the buckling consideration, and a ‘material’ in this paper are based on the findings from the riser tension
(wall, true, or real) tension for calculating tensile stress. The equation verification works.
relates the two is shown in Equation 1. After years of struggling with the riser tension mystery, author
Teff = Tmat + Po Ao − Pi Ai (1) has identified four keys to the puzzle. They are highly inter-related
with each other.
where P is pressure, A is area, subscript o denotes outer, and subscript i 1) Separating the total pressure into hydrostatic head pressure and
denotes inner. Papers [1-4] have been written trying to clarify the applied pressure. It is expressed mathematically by Equation 2.
difference and identify their respective significance. Yet, it only
makes the matter worse. Here are some confusing if not conflicting P = γh + P appl (2)
quotes:
where γ is the weight density (=ρg, ρ is the mass density and g is the
• McIver: Effective Tension – Now you see it, now you don’t. gravity) and h is the water head. Even though both are pressure,
• McIver: Effective tension is not real force; it arises merely for however the hydrostatic head pressure varies with the depth,
convenience of grouping mathematically-alike terms. whereas the applied pressure is constant.
• Sparks: Effective tension is the total force in the pipe/riser column, 2) The mechanics how they affect the riser tension is fundamentally
including contained fluid(s), less the force in the displaced fluid different also; i.e., the hydrostatic head pressure will not generate

1 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
the pressure end cap load (PEL), whereas the applied pressure will. Note the mathematical derivation was for a bent pipe (for
One may loosely classify them as the open-end pipe formula vs. generality), hence, the tension (along the deflected pipe axis) showed
close-end pipe, even though it also depends on the riser shape. up in the lateral equation of motion. If a vertical pipe were used in the
3) Even though riser will deflect due to the external loading, however,
free body diagram, the effective tension term (tension and pressures)
the top-tensioned riser is considered of vertical shape and the
catenary riser is of bent shape. Also, how these two risers are would have no horizontal component. But then, due to the external
tensioned is different; i.e., the vertical top-tensioned riser is by loading, the pipe had to bend. This is what McIver meant “Now you
‘external’ tensioning device, whereas the catenary riser is tensioned see it, now you don’t”.
by its own weight.
4) Whether the riser system is of single string or multiple strings; it is a For a bent pipe subject to internal pressure, the resultant force on
matter of the load path capturing and load distribution among riser the convex side will be greater than the concave side due to the
strings. differential ‘length’ of two sides, as illustrated in Figure 2. Another
unbalanced resultant force will also be generated for the external
Author has been advocating the Multi-Tube riser modeling pressure. These unbalanced resultant forces could initialize an
scheme [8-13] over the conventional Composite model. In the multi- eccentricity needed for buckling. Note both internal and external
tube model, each riser string is represented by a line of elements; pressures are in the effective tension equation, therefore “the effective
whereas the mass, weight, area and moment of inertia are lumped into tension is for buckling consideration” has been stated repeatedly in the
one mathematical line of elements in the composite model. Hence, the riser engineering. This paper will show that these unbalanced
composite model fails to capture the load path correctly, not to resultant forces are nothing but buoyancy, the real potential buckling
mention the interaction between the riser strings. Note the interaction will occur when there is only the internal pressure.
has two meanings: one is the vertical tension distribution among the
strings [9, 11, 12] and the other is the lateral contact load between the Resultant Force
strings [10, 13]. at Concave Side

One important finding out of practicing the multi-tube model was


the Poisson Effect induced tension change, routinely missed in the
composite riser model. For a riser system with concentric pipes and
the inner pipe is tie-backed to the outer pipe at the mudline, as of most Resultant Force
top-tensioned risers, when the outer pipe/inner pipe annulus is at Convex Side

pressurized, there are three tension loads generated. The first one is
the pressure end cap load; the other two are the outer pipe ‘shrinking’ Figure 2 Resultant Forces of a Bent Pipe Subject to Internal Pressure
load and the inner pipe ‘elongating’ load due to the Poisson’s ratio. The term PoAo – PiAi in the effective tension equation is
Note shrinking and elongating loads are used instead of compressive ‘recognized’ as the buoyancy. However, it is so ‘elementary’ that no
and tensile loads; the reason will become apparent with the detailed literature has ever explicitly shown it is indeed the buoyancy for riser
explanation of the driver-reactor theory, which was developed based pipe, yet it is also so ‘complex’ that readers (at least the author) are
on Finite Element Analysis (FEA) results and validated in numerous completely lost in the massive equations derived and presented in the
projects. literatures.

THE SOURCE OF EFFECTIVE TENSION Buoyancy is defined as the differential force between the upward
force acting on the projected horizontal bottom surface of a
The effective tension term appears in the lateral equation of submerged object and the downward force on the projected horizontal
motion for an infinitesimal bent riser pipe (dx), as in Equation 3. The top surface. Looking at Figure 1, it is understandable that a ‘generic’
free body diagram is shown in Figure 1. mathematical expression is very difficult to derive for a bent riser pipe
∂ without any simplification. Seyed and Patel [4] assumed a constant
m&y& + EIy ′′′′ − [(T + Po Ao − Pi Ai ) y ′]′ = Fy (3) curvature pipe in their derivation and used the ‘Fluid Plug’ to explain
∂x the buoyancy.
where m&y& is the inertia, EIy ′′′′ is the flexural, Fy is the lateral
γ w h top A
external load due to current and wave, and (T + PoAo – PiAi) is the
h top γ w h top (A o -A i )
effective tension defined by McIver. In order to make a distinction
between the effective tension (collective terms) and the tension term
(individual T) in the parenthesis, the material (wall, true, or real)
tension was designated to T. Since then, confusions and tug war of
words began. H
y h bot
θ
T γ w (h bot -h top )A γ w (h bot -h top )(A o -A i )
x S
M

Pi
dx
Fy γ w h bot A γ w h bot (A o -A i )
wdx
Po M + ∂M/∂x dx
Figure 3 Buoyancy of a rectangular block and a vertical straight pipe
T + ∂T/∂x dx section in seawater
S + ∂S/∂x dx
θ + ∂θ/∂x dx
Figure 1 Free Body Diagram for a Bent Pipe

2 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
BUOYANCY – SOURCE OF CONFUSION
PEL
To show PoAo – PiAi is indeed the buoyancy for riser pipe, let’s Top Segment Top Segment
start with the simplest geometry, a rectangular block, as shown on the B
Zero
left side of Figure 3. A rectangular block with section area of A and
height of H is submerged in seawater; the top surface is htop from the i+1
th
Segment i+1
th
Segment

mean water line (MWL) and the bottom surface is hbot from MWL B Zero
th
(i.e., H = hbot - htop). Then the buoyancy is γw(hbot – htop)A = γwVol =
th
i Segment i Segment
B Zero
gρwVol, which is exactly the Archimedes principle. i-1
th
Segment i-1
th
Segment
B
If replacing the rectangle block by a vertical straight pipe, as Zero
shown on the right side of Figure 3, the buoyancy will be γw(hbot –
htop)(Ao – Ai). If letting the height H to be infinitesimal (hbot - htop = Bot Segment Bot Segment
B PEL
dx), then the ‘unit’ buoyancy will be γwdxAo – γwdxAi. It is in the
Varyi ng Pressure Constant Pressure
form of PoAo – PiAi, even though so far only the hydrostatic head of Buoyancy, Σ B End Cap Load
seawater is considered. Some important facts about buoyancy of a Differential Load Net PEL
Always Upward Along Axis
vertical straight pipe may be stated based on this elementary
exercise:
Figure 5 Hydrostatic head pressure and applied pressure comparison
1) Force may be caused by mass and acceleration (F = ma) or pressure
and area (F = PA). Dry weight belongs to the former and buoyancy Now let’s turn our attention to buoyancy for a bent pipe. Seyed
is of the latter. Note that Archimedes Principle treats buoyancy as a and Patel had most elaborated derivation of buoyancy for both
body force, due to mass (displaced volume and water mass density) continuous tilted straight and bent pipes by integrating surface
and gravity. Strictly speaking, it is not correct; it is only valid if pressures. They acknowledged that
there is a bottom exposed area.
2) For an in-placed top-tensioned riser, the ‘dry’ makeup hub face at “For a continuous pipe, only the surface area of revolution is subjected to
the subsea wellhead and bottom tieback connector will not be fluid pressure forces and the ends, being continuous, do not interact with
exposed; hence there will be no buoyancy. However, due to this dry the fluid. As a result, the total force experienced by the pipe is not equal
area is relatively small in comparing with the bottom area of tieback to that predicated by the Archimedes principle”
connector, generally it is assumed to be ‘wet’. What they were referring to is primarily for the vertical straight
3) Furthermore, the bottom surface is exposed to the ‘outer’ media as
shown in the left side of Figure 4, which is reprinted from McIver
pipe. It is quite easy to understand it by looking at the right side of
[1]. Note the bottom surface on the right side of figure violates PiAi, Figure 3 if only the surface pressures are integrated without ends.
a misconception committed by author in Reference 7 in which the Then, they went on reinstated the Archimedes principle for continuous
bottom wet surface could be by outer or inner media. Also, with the tilted straight and bent pipes. Figure 6 illustrates the buoyancy of a
illustration of Figure 4, the buoyancy of a vertical straight pipe may bent pipe using the ‘Fluid Plug’.
now be extended to different outer and inner media instead of both
being water. A fluid plug can be of any shape and the resultant load is zero as long as
the fluid is stationary. For an open-end pipe, the ‘balancing’ end cap
load needs to be subtracted for equivalence. Note the PoAo and PiAi are
the fluid forces; the pipe tension will be the ‘reaction’. Also, it is assumed
that the bent pipe is of infinitesimal length; hence, both Po and Pi are of
constant pressure.

Zero
Pi Ai
Pi Pi
= +
Pi Ai

√ × Po

Internal Pressure
Internal Pressure
Fluid Plug
Balancing Load
Figure 4 Definition of ‘Wet’ Surface Po Ao

Zero
4) Buoyancy is the differential force between the bottom surface
upward force and top surface downward force. Therefore, the
location of the vertical straight pipe does not matter and the applied + +
Po
pressure will not change the buoyancy hence the wet weight. The
buoyancy and riser tension mystery is directly caused by letting the Po Ao

height H to be infinitesimal (hbot ≈ htop), the key word “differential” External Pressure External Pressure
Balancing Load
Fluid Plug
is lost in the form of PoAo – PiAi . As the result of it, the top and
bottom pressures turn into ‘side’ pressures. Po Ao

5) The hydrostatic head pressure will not generate pressure end cap
load. This is fairly intuitive to ‘common’ engineers, because never Pi Ai

heard of pressure cap end load associated with hydrostatic pressure. =


Mathematicians may argue otherwise; the reason will be the close- Pi Ai

end pipe formulation is used in the bent pipe buoyancy derivation.


Po Ao
However, because the buoyancy is a differential force, even with the Resultant Buoyancy

close-end pipe formula the hydrostatic head pressure will not affect
the riser tension as the constant pressure. Figure 5 is the comparison Figure 6 Buoyancy of a bent pipe by Fluid Plug
between the hydrostatic head pressure and applied pressure; note the
To verify the buoyancy of a bent pipe, a weightless catenary riser
segments are spaced out for allowing arrows.
was chosen for its shape to study the pressure effect on the riser

3 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
tension. Figure 7 is a magnified deformation plot with the (total weight divided by length). There is no way the buoyancy is a
undeformed shape in black, the deformed shape with constant internal phantom force as suggested by a major oil company’s researcher.
pressure in red and the shape with external pressure in blue. Note only
the constant pressure was applied for now to validate the free body Buoyancy and Pipe Formulas Summary
diagram in Figure 6.
In demonstrating PoAo – PiAi as the riser (pipe) buoyancy; we
started with the simplest vertical straight pipe with hydrostatic head
pressure (varying with depth) and validated Archimedes principle.
Φ s’
Then, we made the length being infinitesimal so that the unit
RFV RFR
buoyancy is in the form of γoAo – γiAi for the vertical pipe. With the
RFH help of fluid plug, we further demonstrated that PoAo – PiAi, as a pair
of constant pressures, is the buoyancy of an infinitesimal bent pipe.
Finally, based on the finite element analysis of a weightless catenary
riser subject to hydrostatic head pressures, we showed ‘total’
θ’
buoyancy is reacted vertically at the hangoff points just as the total
hanging weight (vertically at the hangoff points). Dividing by the arc
y length, the unit buoyancy of a bent pipe with depth varying
hydrostatic head pressure is then proven to be γoAo – γiAi. Note PoAo
To – PiAi and γoAo – γiAi are ‘carefully’ chosen for buoyancy force (with
O x
length) and unit buoyancy, respectively.
Figure 7 Catenary riser deformation (100X) plot Riser pipe in real world must be ‘open’ so that it may serve as the
The axial tension obtained from the verification run was indeed link between the reservoir and surface rig; however, mathematically
PAo for the external pressure case and PAi for the internal pressure there are two pipe formulas, open-end and close-end. There is some
case. Note that the external pressure (blue line) displaced the riser confusion about their applicability. The applications of these two pipe
more than the internal pressure (red line) because Ao is greater than Ai, formulas for a vertical straight or bent pipe subject to depth varying
exactly what the buoyancy is. Figure 6 and Figure 7 answer the (buoyancy) or constant (applied) pressures are summarized in Table 1.
unbalanced resultant forces shown in Figure 2. However, it should be The last row indicates if the pressure end load generated by the
noted that if only the internal pressure is applied, the resultant force on constant pressure is stable or not. And the last column lists the pipe
the convex side may still cause potential buckling. formula should be used.
Table 1 Pipe Formulas Summary
Then, the hydrostatic head pressure (varying with depth) was
applied to the same weightless catenary riser. This is to remove the Vertical Pipe Bent Pipe
infinitesimal assumption made above in demonstrating that PoAo – Pressure To Be Used
PiAi is indeed the buoyancy for a bent pipe. External, internal and Open Close Open Close
combined pressures were modeled by Abaqus’ HPE, HPI and PB Buoyancy No Yes Yes Yes Close
loads, respectively. Note the close-end pipe formulation was chosen
for all three loads. Also, the ‘individual’ external and internal Applied 0 PEL PEL PEL Open with
pressures were studied for mainly the academic purpose and in case CLOAD
Stable Yes Yes Not Yes
their ‘counterpart’ media being of air. The vertical, horizontal and
resultant reactions obtained from the verification analyses are:
Table 1 will be probably the most debated table in this paper, because
RFV = bS (4a) 1) Open-end formulation for the vertical straight pipe has no buoyancy,
which is exactly the case at the bottom tieback connector in a top-
RFH = RFV tan φ = To (4b) tensioned riser. But for the generality of bent pipe, the close-end
pipe formulation has to be used.
2) Riser is designed to be a pressure container; therefore, the close-end
RFR = RFV2 + RFH2 (4c) pipe formulation seems to be the logical choice for the applied
pressure. But for a typical marine drilling riser, the main riser pipe
will never see the pressure end load generated by the internal
where b is the unit buoyancy (γoAo – γiAi), S is the arc length of either pressure contained by the main riser pipe due to the telescope joint.
left side or right side of catenary, Φ is the hang off angle and To is riser Therefore, the open-end pipe with applying PEL as a follower
tension at the catenary bottom (catenary equation’s origin) as shown in concentrated load (CLOAD) shall be used.
Figure 7. It should be noted that the unit buoyancy b is not an 3) Also, if the close-end pipe formulation is used for the applied
input value; it is back calculated from the vertical reaction RFv. constant pressure, the unbalanced resultant force due to the riser
This verification results are documented in Reference 14. curvature, shown in Figure 2, will be missed. Note the close-end
pipe formulation will have zero resultant loads as illustrated by the
Equations 4a to 4c are the same equations for calculating the fluid plug in Figure 6; hence, the close-end pipe formulation is
catenary weight, the only difference is the unit buoyancy is used in categorized as a stable load in Table 1.
Equation 4a instead of unit weight. This is reasonably because the 4) This leads to a practical riser design issue; it has been done routinely
for other design reasons, riser designers probably have no
weight is always acting vertically downward and the buoyancy
knowledge of the implications. For the top-tensioned risers, it is
vertically upward. It is then concluded that both total weight and always a good idea to have the inner riser string (tubing) being the
buoyancy of a catenary riser are ‘accumulated’ vertically along the arc high pressure container so that the unbalanced resultant force caused
length. So, the unit buoyancy for a bent pipe subject to hydrostatic by the riser curvature is contained by centralizers. For risers
head pressure may be obtained by dividing the total buoyancy by the utilizing the outer casing as the high pressure container, such as the
arc length in the same fashion as the unit weight being calculated top-tensioned drilling riser, a higher Tension Factor (to be discussed

4 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
later) is highly recommended. pipe and annulus by approach 1. Five conclusions of significance are
made:
1) If the composite modeling scheme is used for analysis, all three
Too Tight approaches give the identical total mass and effective weight.
Close-end 2) Only the effective weight approach (3) is valid for the multi tube
model which fully captures the load path as illustrated in warning
Apart, Close- messages besides the approach 3’s sub-table.
Too Tight
end and 3) The outer pipe effective weight calculation is independent of
Open-end
Open-end whether there is an inner pipe, and vice verse. The reason is that the
buoyancy is caused by hydrostatic pressure; the riser pipe will only
‘see’ the pressures, whether there is another riser string does not
matter. Hence, the third approach is quite ‘helpful’ in calculating
single pipe and total riser system effective weight.
4) The mud weight may not be carried by riser pipe as suggested by a
major oil company researcher (see Introduction). But the buoyancy
Figure 8 ‘Too tight’ catenary riser deformation (1X) plot
induced by the mud hydrostatic head alters the riser pipe weight; in
5) For catenary risers, generally the outer riser string is the high essence, the riser pipe does carry mud weight. Note the third
pressure container; the riser will just be ‘ballooned’ out without approach does not take the contents into consideration at all, only
buckling instability as along as the hangoff points are sufficiently the hydrostatic head pressures.
apart. Figure 8 is the true scaled deformation plot of the same 5) Effective weight is the sum of dry weight (by mass and gravity) and
catenary riser in Figure 7 with purposely chosen tight hangoff points buoyancy (by pressure and area); so it is a ‘hybrid’ load. When riser
so that the differential length between the convex side and concave effective weight turns into effective tension, to be shown later, the
side is more pronounced. Note the riser remains in catenary shape significance of realizing this becomes apparent.
with the internal pressure using the close-end formulation (blue), but
the shape of the open-end formulation (red) is quite different due to Table 3 is the same three calculations for riser pipes in air (above
the unbalance resultant loads. The deformed shape (black) of the mean water line) by setting the outer media selection key being 0.
catenary riser of Figure 7 is also shown for comparison purpose, Note the effective weight is now the same as dry weight (mass) since
note both the close-end and open-end formulations result in the the riser pipes are now in air. So the same hydrostatic heads
same shape. assumption made in Table 2 calculations is lifted. Also, the tubing
effective weight by approach 3 is the same as that in Table 2; this
EFFECTIVE WEIGHT CALCULATION reinforces the third conclusion above.
With the unit buoyancy firmly established, the unit effective Table 4 is the same three calculations as in Table 2 except that the
weight is written as: heavy mud is in the outer casing/tubing annulus. It is intended to
show the tubing effective weight actually reduced because its ‘outer’
Weff = Wair − (γ o Ao − γ i Ai ) (5a) media (the outer casing/tubing annulus) density increased. Note the
total effective weight is higher just the tubing effective weight is less.
Note the commonly known ‘wet’ weight is a special case of the effect This table echoes the fourth conclusion above, riser carries mud
weight if both γo and γi are of seawater density γw. It maybe written weight.
as:
With the illustration of these three tables, it is concluded that the
Wwet = Wair − (γ w Ao − γ w Ai ) (5b)
effective weight (Equation 5a, Approach 3) is the most universal and
accurate way to calculate the effective weight for both ‘system’ and
Subtracting Equation 5b from Equation 5a and moving Wwet to the individual pipes. It applies to all outer media including air,
right side, particularly ‘powerful’ in capturing the load path.

Weff = Wwet − ((γ o − γ w ) Ao − (γ i − γ w ) Ai ) (5c)


DERIVATION FOLLOWS RISER LOADING HISTORY
Equation 5c has the advantage over Equation 5a if the wet weight data Because the riser effective tension ‘micro’ viewpoint by
is provided, such as marine drilling riser with syntactic foam examining the differential equation led to all kinds of confusions, let’s
buoyancy. take another (macro) approach to derive the effective tension equation
by following the riser loading history. In the real world, the loading
Table 2 shows three approaches to calculate the effective weight
history for a top-tensioned riser system may be divided into the
for a production riser with tubing, assuming both outer and inner
following steps:
hydrostatic heads are the same. Note this assumption will be ‘lifted’
later after explaining Table 3. The first sub-table lists the pipe strings’ 1) Riser joints were made – Dry (Air) Weight
data; the inputs are in the light yellow background. There is a 2) Riser joints were run – Wet Weight
selection key above the outer media density (γo) for riser pipes in air or 3) Tensioner system attached – Tensioned
4) Inner riser string(s) was (were) run – Multi-Tube
in water; it is in water for this set of calculations.
5) Riser in operation – Pressurized
The first approach calculates the pipes and annuli mass and wet 6) Subject to environmental load (Not included in this effective tension
weight separately. The second approach is based on the Archimedes derivation)
principle. The effective weight according to Equation 5a is the third For the full riser length L, hydrostatic head ho for the outer
approach. Note that all three approaches have the same total mass and medium and hi for the inner medium, Equation 5a becomes
‘effective’ weight, but the individual pipe string’s effective weight
calculated by approach 3 is not the same as the sum of wet weight of Weff L = Wair L − (γ o ho Ao − γ i hi Ai ) (6)

5 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Equation 6 takes care of Steps 1 and 2. When the tensioning Full Riser Length Effective Tension Equation Interpretation
device (tensioner or buoyancy can) is attached to the riser system
The significances of looking the effective tension from the
(Step 3), adding tensioner tension Ttnsr to both sides of Equation 6 and
‘Macro’ viewpoint instead of from the ‘Micro’ differential equation
changing the sign due to the weight is acting downward.
are:
Ttnsr − Weff L = Ttnsr − (Wair L − (γ o ho Ao − γ i hi Ai )) (7a) 1) Based on Equation 6, a riser system has its dry weight and effective
weight (wet weight if both outer and inner fluids are seawater)
It maybe rewritten as: depending on the surrounding media. When the tensioning device is
attached, Equations 7a and 7b, the riser weight turns into riser
Ttnsr − Weff L = (Ttnsr − Wair L) + (γ o ho Ao − γ i hi Ai ) (7b) tension. This is the simplest and clearest definition of effective
tension – effective weight plus applied tension.
2) Because the effective weight is a hybrid load, as stated in Item 5 of
When the inner riser strings are inserted into the outer casing Effective Weight Calculation section, the effective tension is then
(Step 4), Equation 6 needs to be ‘multiplied’. Then Equation 7b also a hybrid load. This is the reason why McIver insists “Effective
becomes tension is not real force”. In the mean time, buoyancy will alter riser
weight, therefore riser tension. Sparks’ “Effective tension is simply
Ttnsr − ∑Weff L = (Ttnsr − ∑Wair L) + ∑ (γ o ho Ao − γ i hi Ai ) the real total force in the composite column” is also correct.
3) In determining the tension required to keep the riser upright, only
(8) the effective weight matters as shown in Equation 10a. Note Teff in
the equation is now the riser bottom tension and Ttnsr is the required
The summation sign Σ in Equation 8 signifies that there are multiple
tension. The pressure end cap load can not be depended upon in the
riser strings in the system. required tension calculation for two reasons: one is that the pressure
When the riser system is pressurized (Step 5), the Pressure End may be zero during the storm condition (practical engineering
viewpoint); the other is that it is a part of the material tension not
cap Load (PEL) is generated by the applied pressure, which is the
effective tension (mathematical viewpoint, see Equations 10a and
second component of the total pressure. The pressure end load is 10b).
written as: 4) To further demonstrate that the ‘system’ effective weight is what
matters in the tension requirement calculation, Figure 9 is the free
∑ PEL = ∑ (P i
appl
Ai − P o
appl
Ao ) (9a) body diagram of freestanding tubing used in a production riser
system. There is an expansion joint at the top of the tubing string;
Equation 9a is more generic than the commonly accustomed only hence, the tubing weight is not carried by tensioner. However, the
internal pressure one-term expression. Note for a single string riser, tubing weight will create bending moment which must be offset by
the restoring moment generated by the tensioner tension so that the
the external applied pressure Poappl is zero. Equation 9a may be
riser will stay upright.
rewritten as:

∑ PEL + ∑ (P
Restoring Moment by
o
appl
Ao − Pi appl Ai ) = 0 (9b) Tensioner Tension

Tensioner Tension
Combining Equations 8 and 9b,

Ttnsr − ∑Weff L = [Ttnsr − ∑Wair L + ∑ PEL] + Wave and Current


Load Outer Casing Weight
(10)
[∑ (γ o ho + Poappl ) Ao − ∑ (γ i hi + Pi appl ) Ai ] Tubing Weight

Letting

Teff = Ttnsr − ∑Weff L


Moment due to Outer
(10a) Moment due to Casing Weight
Wave and Current

Tmat = Ttnsr − ∑Wair L + ∑ PEL (10b)


Moment due to
Tubing Weight
Po = γ o ho + P o
appl
(10c)

Pi = γ i hi + Pi appl (10d)

Then, Equation 10 becomes

Teff = Tmat + ∑ Po Ao − ∑ Pi Ai (11)

It is exactly in the form of Equation 1. The difference is that


Equation 1 is for an infinitesimal single riser pipe and Equation 11 is Figure 9 Free Body Diagram of ‘Freestanding’ Tubing
for the full riser system. However, one may limit himself to just one 5) However, the inner riser strings’ effective weights below the mud
riser string and divide Equation 11 by the riser length L to obtain line (the bottom tieback connector, to be precise) should not be
Equation 1. Isn’t that wonderful that another approach from the real included in the required tensioner tension calculation for two
world riser loading sequence point of view matches with the reasons: 1) they will not create moment because no deflection, no
mathematical derivation by free body diagram? moment arm just like the conductor and well casings, and 2) the
‘well-known’ hanging weight through a column vs. weight on the
top of column experiment as shown in Figure 10.

6 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
where RFV (=Weff*S, S is arc length from the origin O to hangoff)
WH is the vertical reaction at the hangoff and s’ is the arc length from
W the hangoff. Both are shown in Figure 7. It should be noted that
WV
Moment the ‘typical’ catenary equation has the arc length s from the origin
W
due to WH to the hangoff, the reason Equation 12a uses s’ (from hangoff) is
prevents Moment
buckling vs. to show RFV acting as the tensioning device Ttnsr as in Equation
due to WV
7a. Also, Equation 12a is for any location along the catenary
Moment
due to W instead of at the bottom of the top-tensioned riser as Equation 7a,
causes because the vertical component will be zero at the bottom of the
buckling
catenary.
W
The horizontal component of the catenary riser axial tension is
Figure 10 Moment by hanging weight vs. weight on top illustration
6) Also, from Figure 9, the restoring moment by the tensioner tension TH = RFV tan φ = To (12b)
needs to resist not only the moment by the riser system effective
weight but also the moment due to external load (wave and current). Note it is constant. The axial tension is the resultant.
Therefore, a Tension Factor (T.F.) is ‘assigned’ to the riser system
for the purpose. Simply because the tension factor is for resisting T = TV2 + TH2 (12c)
the environmental loading, its value varies from field to field and to
be determined iteratively. The higher the tension factor is, the
longer the riser fatigue life will be. But, there is a balance between To be in the effective tension equation form, Equations 12a and
the riser performance and the platform payload. 12b may be rewritten respectively as
7) Once the top tension is determined, because the system effective
weight and the external environmental load are the same regardless Weff ( S − s' ) = (Weff S − Wair s' ) + (γ o Ao − γ i Ai ) s' (13a)
how the tension distributions among riser strings are; the lateral
response will be the same.
Figure 11 is the tension envelope plots for three different tubing
Weff S tan φ = Weff S tan φ (13b)
designs: freestanding tubing on the left, non-preloaded tubing in the
middle and preloaded tieback tubing on the right. Significant Note RFV has been replaced by Weff*S.
different tension distributions among the outer casing and tubing are
observed, however the bottom total effective tensions, defined as the
When the internal pressure is applied to a catenary riser, the
sum of riser strings’ tensions, are the same. For the free-standing pressure end cap load will be generated regardless of the pipe
tubing, the bottom total effective tension is 237.18 kips (= 285.86 – formula (Table 1 Bent Pipe). Even though Poappl is zero because
48.68). For the non-preloaded tubing, it is 239.41 (= 156.83 + the catenary riser is of single bore, for completeness, the pressure
82.58). For the preloaded tubing, it is also 239.41 (= 43.36 + end cap load is written as
196.05). As the result of it, the lateral responses are the same,
witnessed by the same bending moment shown in Figure 12. Note PEL = Pi appl Ai − Poappl Ao (14a)
moments at the bottom and top of the stress joint are identified for
comparison purpose.
It may be rewritten as
This is also the reason the conventional riser composite model, even
though fails to capture the load distribution among the riser strings,
will predict the lateral responses, deflection and bending moment, PEL + Poappl Ao − Pi appl Ai = 0 (14b)
fairly accurately.
8) In author’s opinion, at least for the top-tensioned risers, the Multiplying Equation 14b by sinθ’ to be the vertical component,
commonly called effective tension shall be interpreted as 1) the where θ’ is the slope of the catenary as shown in Figure 7. Note
required tension by the tensioning device to keep the riser upright, Φ and θ’ are complementary angles at the hangoff (s’ = 0).
and 2) the sum of all riser strings’ tensions. The material tension is
Combining the vertical component with Equation 13a,
what the riser analysis is after, to be defined later.

Catenary Riser Tension Derivation Weff ( S − s' ) = [Weff S − Wair s '+ PEL sin θ ' ] +
(15)
Now let’s follow the same ‘macro’ approach for the catenary [(γ o s'+ Poappl sin θ ' ) Ao − (γ i s'+ Pi appl sin θ ' ) Ai ]
riser. Catenary riser differs from the top-tension riser, besides the
shape, in three aspects: Letting
1) Generally, it has only one riser string even though the pipe-in pipe
(PIP) Steel Catenary Riser (SCR) is getting popular in the ultra-deep TeffV = Weff ( S − s ' ) (15a)
water application. Even with PIP SCR, the ‘outer jacket’ is run with
SCR, so the Step 4 of the top-tensioned riser, inserting inner riser
string into outer riser, is not applicable.
V
Tmat = Weff S − Wair s'+ PEL sin θ ' (15b)
2) Catenary riser is tensioned by its own hanging weight, so there is no
tension factor to be calculated. PoV = γ o s'+ Poappl sin θ ' (15c)
3) The catenary riser axial tension has two components: vertical and
horizontal.
PiV = γ i s'+ Pi appl sin θ ' (15d)
Before the riser pressurization, the vertical component at any
location along the riser is Then, Equation 15 becomes
TV = RFV − Weff s' = RFV − (Wair − (γ o Ao − γ i Ai )) s'
TeffV = Tmat
V
+ PoV Ao − PiV Ai (16)
(12a)

7 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
For the horizontal component, multiplying Equation 14b by cosθ’ 1) Pressure end cap load (PEL)
and added to the right side of Equation 13b, 2) ‘Shrinking’ load produced by the ‘Internal’ Pressure on the ‘Outer’
riser string (OR_PI)
Weff S tan φ = [Weff S tan φ + PEL cosθ ' ] + 3) ‘Elongating’ load produced by the ‘External’ Pressure on the ‘Inner’
riser string (IR_PE)
(17)
The tension distribution due to the pressure end load is based on
[( Poappl cosθ ' ) Ao − ( Pi appl cosθ ' ) Ai ] the stiffness ratios of the participating ‘springs’, i.e., riser strings and
Letting tensioner (if any).
The second and third tension loads are due to the Poisson effect,
TeffH = Weff S tan φ (17a) and the tension distribution due to these two loads can be determined
by first identifying which riser string is ‘driving’ the load and which
H
Tmat = Weff S tan φ + PEL cosθ ' (17b) riser string(s) is (are) ‘reacting’ to the load. The ‘driver’ is defined as
the riser pipe that is subject to the annulus pressure. The ‘reactor(s)’ is
(are) the other riser pipe(s) in the parallel spring system.
PoH = Poappl cosθ ' (17c)
The driving axial load is calculated by hoop stress times
Poisson’s ratio times the driver’s cross section area. Each of the
Pi = Pi
H appl
cosθ ' (17d)
reactors carries a load based on its stiffness ratio multiplied by this
axial load, while the driver carries the balancing load needed for
Then, Equation 17 becomes
equilibrium. Note the driver has a load with opposite sign of the
reactors’ loads. The net load is zero; hence, both OR_PI and IR_PE
TeffH = Tmat
H
+ PoH Ao − Pi H Ai (18)
are considered as an internal load.
Even though both the vertical and horizontal components are in The tension load change of each riser string due to one annulus
the form of the effective tension equation with tip-toeing mathematical pressure is the sum of these three sources. The calculation shall be
manipulations and hand waiving, the resultant can never be written in repeated for other annuli’s pressure to get the total tension changes for
the precise form of the effective tension equation. It is ironic that the all pressures.
effective tension originated from the lateral differential equation of a
bent pipe, but it can’t be re-derived for the catenary riser which author Detailed Calculation – Long Chain Form
uses as the bent pipe in demonstrating buoyancy.
A dual casing production riser is chosen for illustrating the
However, in the discussion item 8 of the top-tensioned riser, the calculation. Table 5 shows the riser data, the cells with light yellow
effective tension shall be interpreted for calculating the required background are the inputs. Most of columns are self-explanatory. The
tension by the tensioning device. Since the catenary riser is tensioned ‘Inside’ column shows the geometric relationship of pipes for
by its own weight, there is really no need in dwelling on whether we calculating the annulus area and pressure end load. The ‘Include?’
may re-derive the effective tension equation at all. So the catenary column in the second sub-table signifies whether the riser string
riser axial tension after the riser pressurization is simply Equation 12c stiffness should be included in the calculation, i.e., it is constrained or
with the pressure end cap load. not.
Table 6 has five sub-tables; the first four show the tension
T = TV2 + TH2 + PEL (19)
changes in each of four annuli, the last is the summation. In the first
sub-table, the pressure end load is distributed to each spring per
Note there is no subtraction of PiAi in Equation 19. stiffness ratio. The OR_PI load (39.91) is calculated as
Seyed and Patel [4] emphasized in their paper about the flowing [500*(12.347/2)/0.514]*0.32* 20.77/1000, where 500 is the pressure
fluid effect on the catenary riser material tension. The steady flow in psi, 12.347” is the outer casing ID, 0.514” is the outer casing
effect on the riser dynamic is less important than the slug flow, and the thickness, 0.32 is the Poisson’s ratio, 20.77 is the outer casing area in
top-tensioned riser is less affected by the slug flow than the highly in2 and 1000 is to convert lbs to kips. Due to the Poisson’s effect, the
curved riser, such as Lazy Wave. There are also extensive discussions Outer riser string (outer casing in this case) is shrinking; hence, it is
about the issue in DNV-OS-F201 Appendix A, Section H; echoing the with a negative sign and reactors are in compression. But the driver
fact “Steady flow only affects the material tension in the same manner itself, with double line boarder, is actually in tension to balance the
to internal pressure and leaving the effective tension unchanged”. The reactors’ compressions.
difference between the effective tension and the material tension is The IR_PE load (21.96) is calculated as [500*(9.625/2)/0.545]
becoming clear – pressure end load and flow-induced load. *0.32* 15.55/1000, where 9.625” is the inner casing OD, 0.545” is the
inner casing thickness and 15.55 in2 is the inner casing area since the
DRIVER - REACTOR THEORY driver now is the inner casing. The Inner riser string (inner casing) is
elongating due to Poisson’s effect, it stretches all reactors but itself is
Finally, author will like to present the ‘MORE’ part of the paper, in compression. The tubing and gas lift line are not affected by the
the tension changes due to Poisson Effect. It was ‘discovered’ in the pressure in outer casing/inner casing annulus; hence, these two
Pressure Test step in the Riser Installation/Completion Sequence columns are ‘grayed’ out. The last column is the sum of PEL, OR_PI,
Simulation Analysis [9]. and IR_PE, the tension change in each spring due to 500 psi pressure
in this annulus.
For a concentric pipe-in-pipe riser system and the inner riser
string(s) is (are) ‘constrained’, when the pressure is applied to the The second sub-table is for the inner casing annulus pressure.
riser annulus, there are three sources of tension loads generated in Since both tubing and gas lift line are inside the inner casing (see
addition to the existing tension, as shown in Figure 13. ‘Inside’ column of Table 5), therefore, they are both of IR_PE. The

8 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
third sub-table is for the tubing annulus pressure and the fourth sub- concrete to riser pipe and its content, and the other was the direct
table is for the gas lift line. Both sub-tables have only one OR_PI; the equivalence of deviator stress to effective stress. Since the hydrostatic
calculations are under the proper column. stress is the second component of the total stress, then its counterpart
in the tension equation will be the buoyancy. As a result of it, he was
The fifth sub-table recaps the first four calculations in the 2nd to
able to have a meaningful physical interpretation of effective tension
5th columns; they are color coded for ease of comprehension. The
and successfully demonstrated von Mises calculations by both
total tension change in each spring is the sum, listed in the next
material and effective tensions.
column. The last column of the table is the results from ABAQUS.
The correlation is excellent. Using effective tension to calculate axial stress is perfectly fine
before riser is pressurized. It can even be used for a single string riser
Alternative Calculation – Short Sweet Form with internal applied pressure, i.e., catenary riser with pressure end
load. However, the effective tension will not be affected by the flow
The driver – reactor theory was originally developed to calculate induced tension and Poisson ratio induced loads, material tension has
the tension change for one annulus pressure. It was later ‘easily’ to be used in the stress calculation. If the structural integrity check is
extended to multiple annuli pressures, yet the criticism has been to its load-based instead of stress-based, the material tension shall be used
lengthiness and complexity. Alternative driver equilibrium force for the same token.
calculation was then developed and illustrated in Table 7 for OR_PI
and IR_PE of the first annulus pressure.
First distribute Force due to Poisson effect (F_PSN, either OR_PI CONCLUSIONS
or IR_PE) to each ‘spring’ per stiffness ratio, then the Equilibrium This paper started with the conflicting quotes regarding the riser
Force (F_EQU, equals to F_PSN with a reversed sign) is added to the tension from four most authoritative papers [1-4]. The infamous
driver. The beauty of this approach is that F_PSN may be effective tension was re-derived following the riser load history to
accumulated for all OR_PI’s and IR_PE’s in much more compact interpret the equation from the ‘global’ viewpoint. Four keys to the
form, shown in Table 8. riser tension puzzle were identified and utilized along the re-
derivation. The old-school riser engineering gospel “The effective
In Table 8, the spring and its stiffness ratio are listed in the first
tension for buckling consideration and the material tension for stress
and second column, respectively. Both Pressure End Load (PEL in the
calculation” is re-emphasized with slight modifications, particularly
3rd column) and Force due to Poisson effect (F_PSN in the 4th column)
with a clear division of before and after the riser pressurization.
are distributed per the stiffness ratio. The driver equilibrium force
(F_EQU in the 5th column) is then ‘subtracted’. The total tension Driver – Reactor Theory, both Long Chain and Short Sweet
change in each spring for all annuli pressures are listed in the forms, were presented to calculate the loads induced by the Poisson
‘Combined’ column. They are identical to the detailed calculation. effect for a concentric pipe-in-pipe riser system and constrained inner
riser strings. The results reveal significantly different load
The alternative calculation is called by author as the Short Sweet
distributions from the calculation without considering the driver
form (in memory of Dr. Harry Sweet) of the Driver – Reactor theory
equilibrium load.
and the detailed calculation is the Long Chain form.
Table 9 is the riser tension clarification summary. It follows the
One very important fact is that the loads due to Poisson effect
riser operation sequences. The equation number in this paper is listed
will only be generated if the inner riser string(s) is (are) constrained.
in the second column. The next three columns are the Left Side Term,
Otherwise, the driver may shrink or elongate but no load will be
Right Side First Term, and Right Side Second Term; one may loosely
induced. Simply because it is of ‘displacement’ limited problem, the
classify as Effective Term, Material Term and Buoyancy, respectively.
driver Equilibrium load always comes with the load due to Poisson
The last column lists the physical quantity to be used for stress
effect. Calculation assuming the Poisson induced load follows
calculation. Note for the first two stages, the material tension may
stiffness ratio without the equilibrium load is simply not correct (see
also be used even though the effective tension is listed.
the last isolated column in Table 8).
The “Multi-Tube Tension” at the Inner Strings Added stage
It seems that the discussion of riser strings’ tension changes is
means the individual ‘material’ tension in each riser string depending
limited to the applied pressure only. What about the Poisson’s effect
on the ‘boundary condition’ of the inner strings, i.e., tie-back without
due to the depth varying pressure (buoyancy)? The answer is none.
preload, tie-back with preload, free-hanging, or free-standing. The
The reason is that Poisson induced load will only occur if the pipe is
“MT Tension with Driver – Reactor” at the last cell is for riser systems
constrained. For both the top-tensioned riser and catenary riser, the
with constrained inner strings.
riser is free-hanging during deployment. Hence, no load will be
generated. In summary, the key messages that this paper is trying to deliver
are listed in the following one-liners:
The Driver – Reactor theory has been validated in several top-
tensioned riser projects the author was involved in. It is also • The simplest definition of effective tension is the riser system
applicable for the Pipe-in-Pipe Steel Catenary Riser; one example was effective weight plus the applied tension; which may be provided by
the BP Atlantis project. Author also found it helpful in understanding external tensioning device or riser self weight. The effective tension
the tension slope of a Load Sharing Marine Drilling Riser [12], even remains constant once the riser system is tensioned.
though the auxiliary lines were not concentric, hence only OR_PI • The material tension is the tensile load integrated from linearized
membrane stress; a ‘reversed’ tension - stress definition but the most
loads were considered.
accurate. It varies with flow induced load, pressure end cap load
and Poisson ratio induced loads.
• Separating the total pressure into hydrostatic head pressure and
STRESS CALCULATION applied pressure is the most important key among the four identified
Sparks [3] applied two Civil Engineering concepts to Riser in the paper to understand the difference between the effective
Engineering; one was the composite column analogy of reinforced tension and material tension.

9 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
• Hydrostatic head pressure (Buoyancy) will not generate the pressure from tension and moment, the material tension should be used. If
end cap load but the applied pressure will. the Lamé formulation [3] is chosen for the von Mises stress
• Buoyancy is the differential load between the bottom surface calculation, the total pressures should be used in converting the
upward force and top surface downward force; the location of a material tension to effective tension.
submerged object does not matter and the applied pressure will not 4) If the composite model was used in the analysis, the tension changes
change buoyancy. due to Poisson effect can be ‘easily’ included in the post-processor,
• The buoyancy and riser tension mystery is directly caused by letting particularly the Short Sweet form.
the height H to be infinitesimal; the key word “differential” is lost
Here are some recommendations for riser designers:
in the form of PoAo – PiAi and turns top and bottom pressures into
‘side’ pressure. Of course, mixing the hydrostatic head pressure 1) Ask riser analysts about whether the pressure end cap load is
with applied pressure makes the matter worse. included in the riser tension data provided for the component design
• Buoyancy will always come into play because the riser will always and detail finite element analysis.
be deflected due to the external loading except a very small dry hub 2) For the top-tensioned risers, use the inner riser string to be the high
face area between the subsea wellhead and the bottom connector, if pressure container. If the outer riser string has to be used as the high
the riser system is connected. pressure container, high tension factor shall be ‘assigned’ to avoid
• For a deploying riser, the whole riser system will be buoyed. any potential instability.
• Effective weight equation is the most robust formula for riser system 3) For the catenary risers, keep the ‘nominal’ hangoff angle to be 8 to
and individual string weight calculation; applicable for all media 15 degrees to not only meet the bend radius limit but also avoid the
including air and most powerful in capturing the load path. instability.
• Wet weight is a special case of the effective weight.
• Effective weight is a ‘hybrid’ load; the sum of gravity induced load
(dry weight) and pressure induced load (buoyancy).
ACKNOWLEDGMENTS
• In calculating the required tension to keep the riser system upright, Author is honored to have Professor M.H., Patel, and Dr. Pieter
only the system effective weight above the mudline matters. Wybro reviewing the paper and providing their viewpoints.
• The pressure end cap load should not be accounted for in the
required tension calculation.
• Even if an inner riser string weight is not physically carried by the REFERENCES
tensioner, e.g., freestanding tubing, its weight should be included in [1] McIver, D.B. and Olson, R.J., “Riser Effective Tension – Now You
the system effective weight calculation. See It, Now You Don’t”, 37th Petroleum Mechanical Engineering
• For a multiple strings riser system, the effective tension is the sum of Workshop and Conference, Dallas, September 1981, pp 177-187.
the material tensions of all riser strings. [2] McIver, D.B., “The Effect of Hydrostatic Pressure on Pipes (The
• As long as the total effective tension is the same, the lateral Concept of Effective Tension)”, source unknown.
responses (displacement and bending moment) will be the same [3] Sparks, C.P., “The Influence of Tension, Pressure and Weight on
even if the individual riser string tensions are different. Pipe and Riser Deformations and Stresses”, Transactions of ASME,
• This is the reason that the composite model predicts the bending Vol. 106, March 1984, pp 46 – 54.
moment quite accurately even though it fails to capture the load [4] Seyed, F.B. and Patel, M.H., “Mathematics of Flexible Risers
path. Including Pressure and Internal Flow Effects”, Marine Structures 5,
• Effective tension for buckling is true for top-tensioned risers, if the Elsevier Science Publishers Ltd., 1992, pp 121 – 150.
word buckling is defined as falling on the seafloor. Just simply need [5] API RP 2RD, Design of Risers for Floating Production Systems
to have sufficient tension, with tension factor, to resist the moments (FPSs) and Tension-Leg Platforms (TLPs), 1st Edition, June 1998.
by the system effective weight and environmental loads. [6] DNV-OS-F201, “Dynamic Risers”, 2001
• The ‘real’ instability, commonly known as buckling, is due to the [7] Chang, R., “Using ABAQUS to Verify Riser Tension”, ABAQUS
unbalanced resultant forces between the convex side and concave User Conference, 2000.
side when the internal pressure is applied. [8] Chang, R., Fisher, E., Underland, H., and Wu, M., “Riser Modeling
• For catenary risers, the effective tension has no ‘practical’ physical Scheme: Multi-Tube vs. Composite” ASME/OMAE 2000, 8012.
meaning because it is tensioned by self weight with probably some [9] Chang, R. and Smedley, M., “Multi-Tube Model Application In Riser
added weights for SCR. Installation/Completion Simulation Analysis” ASME/OMAE 2003,
• Never under-estimate the Poisson induced loads when the inner riser 37416.
strings are constrained. [10] Chang, R. and Yu, J., “Multi-Tube Model Application In Riser
• Driver – Reactor theory is applicable for the applied pressure only Centralizer Analysis” ASME/OMAE 2003, 37417.
and the equilibrium load (F_EQU) must be considered for the [11] Chang, R. and Fisher, E., “Riser Modeling Scheme for a Load-
Poisson induced load (F_PSN) to be an internal load. Sharing Marine Drilling Riser”, ASME/OMAE 2007, 29010.
• Both McIver and Sparks are right, they are just referring to the [12] Chang, R. and Fisher, E., “Load-Sharing Marine Drilling Riser Pipe
different stages of riser loading history; Sparks is before the riser Tension Slope”, ASME/OMAE 2007, 29011.
system pressurization and McIver is after. [13] Chang, R., “Drilling Induced Vibration May Cause Riser Fatigue –
Modifying Drilling Practices can Avoid Damage Related to Close
Some suggestions and words of precaution for riser analysts: Annular Tolerances”, World Oil, October, 2007, pp 67 – 72.
[14] “Using Catenary Riser to Verify A Bent Pipe Buoyancy”, ERC
1) Practice the multi-tube modeling scheme whenever possible within Internal Report.
the limitation of available riser program.
2) Analysis procedure shall follow the riser operation as close as
possible to have more realistic results.
3) In post-processing analysis results, i.e., calculating the axial stress

10 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 2 Effective Weight Calculation for Production Mode Content
Riser String Data In Air (0) or in Water (1) 1
Riser OD Thickness ID Ao Ai Area W_air W_wet Annulus Content γo γi
Pipe (in) (in) (in) (in2) (in2) (in2) (lbs/in) (lbs/in) (in2) (PPG)
3
(lbs/in ) (lbs/in3)
Outer 10.75 0.485 9.78 90.76 75.12 15.64 4.440 3.860 36.64 2.8 0.037 0.012
Tubing 7 0.5 6.00 38.48 28.27 10.21 2.898 2.520 28.27 8.23 0.012 0.036
Total Wet Weight 6.380

Approach 1: Calculate Pipes and Annuli Mass and 'Wet Weight' Separately
Mass W_wet ΣMass ΣW_wet
(lbs/in) (lbs/in) (lbs/in) (lbs/in)
Pipe1 4.440 3.860 Pipe1 + Annu1 4.884 2.948
Pipe2 2.898 2.520 Pipe2 + Annu2 3.906 2.480
Annu1 0.444 -0.913 TOTAL 8.789 5.428
Annu2 1.007 -0.040
TOTAL 8.789 5.428

Approach 2: Archimedes Principle, Total Dry Weight (Mass) - Total Displaced Volume by Outer Media
Total Dry Weight 8.789 (lbs/in)
Displaced Outer Media 3.362 (lbs/in)
5.428

Approach 3: Effective Weight, W_air - (γoAo-γiAi) for each tube


Mass Weff
(lbs/in) (lbs/in)
Multi-tube 1 4.884 1.989 Not equal to 2.948, which is the sum of Pipe1 (3.86) and Annu1 (-0.913) in Aprroach 1
Multi-tube 2 3.906 3.439 Not equal to 2.48, which is the sum of Pipe2 (2.52) and Annu2 (-0.04) in Aprroach 1
TOTAL 8.789 5.428

Table 3 Effective Weight Calculation for Pipes in Air


Riser String Data In Air (0) or in Water (1) 0
Riser OD Thickness ID Ao Ai Area W_air W_wet Annulus Content γo γi
Pipe (in) (in) (in) (in2) (in2) (in2) (lbs/in) (lbs/in) (in2) (PPG) (lbs/in3) (lbs/in3)
Outer 10.75 0.485 9.78 90.76 75.12 15.64 4.440 3.860 36.64 2.8 0.000 0.012
Tubing 7 0.5 6.00 38.48 28.27 10.21 2.898 2.520 28.27 8.23 0.012 0.036
Total Wet Weight 6.380

Approach 1: Calculate Pipes and Annuli Mass and 'Wet Weight' Separately
Mass W_wet ΣMass ΣW_wet
(lbs/in) (lbs/in) (lbs/in) (lbs/in)
Pipe1 4.440 4.440 Pipe1 + Annu1 4.884 4.884
Pipe2 2.898 2.898 Pipe2 + Annu2 3.906 3.906
Annu1 0.444 0.444 TOTAL 8.789 8.789
Annu2 1.007 1.007
TOTAL 8.789 8.789

Approach 2: Archimedes Principle, Total Dry Weight (Mass) - Total Displaced Volume by Outer Media
Total Dry Weight 8.789 (lbs/in)
Displaced Outer Media 0.000 (lbs/in)
8.789

Approach 3: Effective Weight, W_air - (γoAo-γiAi) for each tube


Mass Weff
(lbs/in) (lbs/in)
Multi-tube 1 4.884 5.350 Not equal to 4.884, which is the sum of Pipe1 (4.44) and Annu1 (0.444) in Aprroach 1
Multi-tube 2 3.906 3.439 Not equal to 3.906, which is the sum of Pipe2 (2.898) and Annu2 (1.007) in Aprroach 1
TOTAL 8.789 8.789

Table 4 Effective Weight Calculation for Heavy Mud in Outer Annulus


Riser String Data In Air (0) or in Water (1) 1
Riser OD Thickness ID Ao Ai Area W_air W_wet Annulus Content γo γi
Pipe (in) (in) (in) (in2) (in2) (in2) (lbs/in) (lbs/in) (in2) (PPG)
3
(lbs/in ) (lbs/in3)
Outer 10.75 0.485 9.78 90.76 75.12 15.64 4.440 3.860 36.64 10 0.037 0.043
Tubing 7 0.5 6.00 38.48 28.27 10.21 2.898 2.520 28.27 8.23 0.043 0.036
Total Wet Weight 6.380

Approach 1: Calculate Pipes and Annuli Mass and 'Wet Weight' Separately
Mass W_wet ΣMass ΣW_wet
(lbs/in) (lbs/in) (lbs/in) (lbs/in)
Pipe1 4.440 3.860 Pipe1 + Annu1 6.026 4.090
Pipe2 2.898 2.520 Pipe2 + Annu2 3.906 2.480
Annu1 1.586 0.229 TOTAL 9.931 6.570
Annu2 1.007 -0.040
TOTAL 9.931 6.570

Approach 2: Archimedes Principle, Total Dry Weight (Mass) - Total Displaced Volume by Outer Media
Total Dry Weight 9.931 (lbs/in)
Displaced Outer Media 3.362 (lbs/in)
6.570

Approach 3: Effective Weight, W_air - (γoAo-γiAi) for each tube


Mass Weff
(lbs/in) (lbs/in)
Multi-tube 1 6.026 4.330 Not equal to 4.09, which is the sum of Pipe1 (3.86) and Annu1 (0.229) in Aprroach 1
Multi-tube 2 3.906 2.240 Not equal to 2.48, which is the sum of Pipe2 (2.52) and Annu2 (-0.04) in Aprroach 1
TOTAL 9.931 6.570

11 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Production Riser Tension Envelope Production Riser Tension Envelope Production Riser Tension Envelope
Free-Standing Tubing Tubing with TB Hanger (Un-Preloaded) Tubing with FLX Hanger (Preloaded)
1200 1200 1200

MWL
MWL MWL
1000 1000 1000

800 800 800


Elevation (ft)

Elevation (ft)

Elevation (ft)
Riser Riser Riser
600 600 600
Tubing Tubing Tubing

400 400 400

Bottom Effective Tension Bottom Effective Tension Bottom Effective Tension


= 237.18 = 285.86 + -48.68 = 239.41 = 156.83 + 82.58 = 239.41 = 43.36 + 196.05

200 200 200

Stress Joint Stress Joint Stress Joint


-48.68 285.86 82.58 156.83 43.36 196.05
0 0 0
-100 0 100 200 300 400 -50 0 50 100 150 200 250 -100 0 100 200 300
Tension (kips) Tension (kips) Tension (kips)

Figure 11 Tension Comparison Plot

Production Riser Moment Envelope Production Riser Moment Envelope Production Riser Moment Envelope
Free-Standing Tubing Tubing with TB Hanger (Un-Preloaded) Tubing with FLX Hanger (Preloaded)
1200 1200 1200

MWL MWL MWL


1000 1000 1000

800 800 800


Elevation (ft)
Elevation (ft)

Elevation (ft)

600 Riser 600 Riser 600 Riser


Tubing Tubing Tubing

400 400 400

200 200 200

Stress Joint Stress Joint Stress Joint


81 80 79

1180 1181 1199


0 0 0
0 250 500 750 1000 1250 0 250 500 750 1000 1250 0 250 500 750 1000 1250
Moment (ft-kips) Moment (ft-kips) Moment (ft-kips)

Figure 12 Bending Moment Comparison Plot

12 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
PEL

OR_PI

Equilibrium

D R

IR_PE
Equilibrium

R D R

Figure 13 Three Loads Generated When Annulus Is Pressurized

Table 5 Riser Data for Driver – Reactor Theory


TENSION CHANGES DUE TO PRESSURE & POISSON RATIO
Riser String Data
Riser OD Thickness ID Ao Ai Area Annulus Pressure PEL
Name Inside
Pipe (in) (in) (in) (in2) (in2) (in2) (in2) (psi) (kips)
1 Outer 13.375 0.514 12.347 140.50 119.73 20.77 46.97 500 23.49
2 Inner 9.625 0.545 8.535 72.76 57.21 15.55 1 38.47 500 19.24
3 Tubing 4.5 0.271 3.958 15.90 12.30 3.60 2 12.30 6000 73.82
4 Gas Lift 1.9 0.145 1.61 2.84 2.04 0.80 2 2.04 300 0.61
117.16
Spring Name Stiffness Include? Active Ratio * Poisson Ratio 0.32
1 Outer 137 1 137 0.408
2 Inner 101 1 101 0.301
3 Tubing 23 1 23 0.068
4 Gas Lift 5 1 5 0.015
5 Tensioner 70 1 70 0.208
(kips/ft) 336 1.000

13 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 6 Detailed Calculation of Tension Changes

Driver - Reactor Theory, Detailed Calculation


Outer Inner Tubing Gas Lift
For Outer String, Pressure of 500 (psi)
PEL OR_PI IR_PE
SPRING Combined
23.49 -39.91 21.96
Outer 9.58 23.64 8.96 42.17
Inner 7.06 -12.00 -15.36 -20.30
Tubing 1.61 -2.73 1.50 0.38
Gas Lift 0.35 -0.59 0.33 0.08
Tensioner 4.89 -8.31 4.58 1.15
Sum 23.49 0.00 0.00 23.49
For Inner String, Pressure of 500 (psi)
PEL OR_PI IR_PE IR_PE
SPRING Combined
19.24 -19.48 4.78 0.84
Outer 7.84 -7.94 1.95 0.34 2.19
Inner 5.78 13.62 1.44 0.25 21.09
Tubing 1.32 -1.33 -4.46 0.06 -4.41
Gas Lift 0.29 -0.29 0.07 -0.83 -0.76
Tensioner 4.01 -4.06 1.00 0.17 1.12
Sum 19.24 0.00 0.00 0.00 19.24
For Tubing String, Pressure of 6000 (psi)
PEL OR_PI
SPRING Combined
73.82 -50.48
Outer 30.10 -20.58 9.52
Inner 22.19 -15.17 7.02
Tubing 5.05 47.03 52.08
Gas Lift 1.10 -0.75 0.35
Tensioner 15.38 -10.52 4.86
Sum 73.82 0.00 73.82
For Gas Lift String, Pressure of 300 (psi)
PEL OR_PI
SPRING Combined
0.61 -0.43
Outer 0.25 -0.17 0.08
Inner 0.18 -0.13 0.06
Tubing 0.04 -0.03 0.01
Gas Lift 0.01 0.42 0.43
Tensioner 0.13 -0.09 0.04
Sum 0.61 0.00 0.61
Note: All loads are in kips

Tension Change Summary


Due to Due to Due to Due to Total ABAQUS
SPRING Outer Inner Tubing Gas Lift Change Result
500 (psi) 500 (psi) 6000(psi) 300 (psi) (kips) (kips)
Outer 42.17 2.19 9.52 0.08 53.96 53.90
Inner -20.30 21.09 7.02 0.06 7.87 7.80
Tubing 0.38 -4.41 52.08 0.01 48.06 48.00
Gas Lift 0.08 -0.76 0.35 0.43 0.10 0.10
Tensioner 1.15 1.12 4.86 0.04 7.18 7.20
Sum 23.49 19.24 73.82 0.61 117.16 117.00

14 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Table 7 Alternative Calculation for Individual Load

OR_PI F_PSN F_EQU


-39.91 -39.91
23.64 ═ -16.27 39.91
-12.00 -12.00
-2.73 -2.73
-0.59 -0.59
-8.31 -8.31
0.00 -39.91 39.91 0.00

IR_PE F_PSN F_EQU


21.96 21.96
8.96 8.96
-15.36 ═ 6.60 -21.96
1.50 1.50
0.33 0.33
4.58 4.58
0.00 21.96 -21.96 0.00

Table 8 Alternative Calculation of Tension Changes

Alternative Calculation (Short Sweet Form) Wrong Way


Stiffness PEL F_PSN F_EQU PEL+F_PSN
SPRING Combined
Ratio 117.16 -82.71 34.45
Outer 0.408 47.77 -33.72 39.91 53.96 14.05
Inner 0.301 35.22 -24.86 -2.49 7.87 10.35
Tubing 0.068 8.02 -5.66 45.70 48.06 2.36
Gas Lift 0.015 1.74 -1.23 -0.41 0.10 0.51
Tensioner 0.208 24.41 -17.23 7.18 7.18
Sum 1.000 117.16 -82.71 82.71 117.16 34.45

Note: PEL (117.16) distribute to each 'spring' per stiffness ratio


F_PSN (-82.71) = -39.91 + 21.96 + -19.48 + 4.78 + 0.84 + -50.48 + -0.43
then distribute to each 'spring' per stiffness ratio
F_EQU for Outer (39.91) = -(-39.91 + 0 + 0 + 0 )
F_EQU for Inner (-2.49) = -(21.96 + -19.48 + 0 + 0 )
F_EQU for Tubing (45.7) = -(0 + 4.78 + -50.48 + 0 )
F_EQU for Gas Lift (-0.41) = -(0 + 0.84 + 0 + -0.43 )
F_PSN (-82.71) may also be the negative of Sum of F_EQU (82.71)

Table 9 Riser Tension Clarification Summary

Stress
Stage Equation L.S. Term R.S. 1st Term R.S. 2nd Term
Calculation
Effective (Wet) Air (Dry) Weight Buoyancy Effective Weight
Deployment 6
Weight (WeffL) (WairL) (γohoAo – γihiAi) (WeffL)
Effective Tension Material Tension Buoyancy Effective Tension
Tensioned 7
(Ttnsr – WeffL) (Ttnsr – WairL) (γohoAo – γihiAi) (Ttnsr – WeffL)
Inner Strings Effective Tension Material Tension Buoyancy Multi-Tube
8
Added (Ttnsr – ΣWeffL) (Ttnsr – ΣWairL) Σ(γohoAo – γihiAi) Tension
Effective Tension Material Tension Buoyancy MT Tension with
Pressurized 10
(Ttnsr – ΣWeffL) (Ttnsr – ΣWairL + ΣPEL) Σ(PoAo – PiAi) Driver - Reactor

15 Copyright © 2009 by ASME

Downloaded 10 Oct 2011 to 212.140.243.34. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

You might also like