SPE 90151

The Heat Transfer Characteristics of Vacuum Insulated Tubing

J. H. Azzola, SPE, Thermal Science Inc.; P. D. Pattillo, SPE, BP America; J. F. Richey, SPE, BP America; S. J. Segreto,
SPE, Tube-Alloy Products

Copyright 2004, Society of Petroleum Engineers Inc.

exterior coupling insulators, and using high-temperature steam
This paper was prepared for presentation at the SPE Annual Technical Conference and injection.
Exhibition held in Houston, Texas, U.S.A., 26–29 September 2004.

This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
Introduction
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to Over the past 25 years, VIT has matured as a technology
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at capable of thermal isolation in increasingly challenging
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
environments1-4. Historically, because VIT performance in
for commercial purposes without the written consent of the Society of Petroleum Engineers is some applications, such as cold startup and permafrost
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous subsidence prevention, proved adequate, design tools have
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. remained undeveloped. In other applications, such as paraffin
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
formation and annular pressure buildup (APB), insulated
tubing design requires a deep understanding of the
Abstract performance boundaries in order to make rational cost/benefit
Thermal effects in vacuum insulated tubing (VIT) have been judgments. In some notable cases (steam flood and heavy oil
studied in previously reported experiments2. VIT is usually flow applications), such understanding is necessary to
characterized by a single thermal conductivity ( k -value) determine whether the economic exploitation of a field is
based on a one-dimensional radial heat transfer analogy. feasible at all. Unfortunately, the inherent complexity of the
Despite its simplicity, however, this physical model masks the subject does not lend itself to simple answers. Nevertheless,
underlying complexity of the two-dimensional axisymmetric with regard to VIT and insulated tubing in general,
heat transfer field. establishing performance, while far from being simple, is not
Recent theoretical insights have demonstrated that VIT an altogether intractable problem.
behaves as a fin around the weld and coupling regions, where VIT performs well for tasks that require a combination of
up to 90% of the overall heat loss can occur. As a result, VIT 1) compact space such as those encountered downhole and in
does not have a single k -value; instead it has a potentially riser sections, 2) high load capability, able to pull 200,000 -
wide range of k -values, depending on boundary conditions. 500,000 lbs tension load, and 3) high thermal insulation value,
Recent numerical and experimental investigations of vertically with overall thermal conductivity ( U -values) less than 1
aligned VIT demonstrate the highly sensitive nature of thermal Btu/hr-ft2-°F (5.7 W/m2-°K).
performance with respect to these boundary conditions1. With This paper addresses issue (3), concentrating on the
this insight, previous test data2 is reinterpreted in a more thermal performance of VIT under as wide a range of
revealing light. It can be demonstrated that the 10-fold spread boundary conditions as is presently encountered in the
of k -values from these tests is attributable to explainable industrial use of the product. Conceptual issues about
boundary phenomena. performance will be addressed. To standardize requirements
Another issue which requires clarification is the k -value for all users, a choice of metric is presented, together with a
definition. A standard definition is presented here which 1) is rationale for its use.
physically related to the effective thermal conductivity of the
vacuum and 2) is relatively stable for different VIT The Physics of VIT
configurations. To understand conceptually how VIT works, one must start
Transient behavior, including startup, cool down during with the vacuum insulation jacket. Thermal insulation from
shut in, ground soak, and convection cell development will be the vacuum body is extremely high. An alternate
explained, as well as the implications for long-term thermal interpretation is that the heat flux (heat flow per unit area)
stability. through the vacuum space is relatively low. The complicating
The discussion concludes with calculations of k -value for feature of VIT is the relatively large heat flux in all directions
VIT under significantly varied boundary conditions: different around the coupler1. There are three distinct heat paths, each
annular fluids, different soils, with and without interior or preferentially following low thermal resistance through the
relatively high conductivity steel. The first path runs radially
2 SPE 90151

through the vacuum space ( q1 in Fig. 1). The second path That this conceptualization is inaccurate is subtle, for it
runs radially through the coupler ( q 2 in Fig. 1). The third requires an understanding of the interactions between the heat
flowing through the weld and coupler with the often complex
path runs axially along the inner pipe toward the weld, then fluid flow patterns that develop in the annular spaces of
through the weld and axially down the outer pipe, in a hot typical wellbore completions. It is because of these
zone from one to three feet beyond the weld. In this hot zone, interactions that VIT must be analyzed as an integral unit, as
heat is lost by convection ( q 3 in Fig. 1). will be demonstrated below.

The term q 3 considerably complicates a quantitative

Heat flow through
weld and axially in estimate of total heat loss, for the size of the hot zone around
weld region is 50 - 90% of heat loss
significant is around coupler the weld is directly a function of amount of convective heat
and weld zones loss around this zone. In fact, q 3 can be 30 – 70% of total
Coupler heat loss (Fig. 2). Convective heat loss is affected by the
magnitude of the buoyancy induced flow pattern, which is
driven by the temperature profile in the weld region. Physical
Heat path Weld
prediction is further complicated by the fact that the
q2 Hot zone around convection pattern in the surrounding fluid media in the
Coupler extends
approx. 1 – 2 feet annulus around the VIT exhibits a fully two-dimensional,
Heat axially from coupler sometimes time-dependent behavior. For these reasons, the
path centerline
heat transfer problem is fully coupled and boundary condition
q3
dependent. There is no one insulation value ( k -value, U -
value) associated with VIT. Instead, there is a fifty-fold range
Vacuum Heat flux around body (q1) is of values depending on boundary conditions (annular fluid
space approx. 10 – 50% of total VIT
heat loss
properties, completion geometry, coupling insulation, ground
conductivity, etc.)1.
To understand the fin analogy of heat loss q 3 , consider the
Inner Pipe Outer Pipe
heat equation in steady state:
( )
Q = T1 − T2 .. .......................................................... (7)
Fig. 1 –Heat flow (arrows) around VIT coupler region and body. R
For VIT, there is a relation between R and k -value:
∆T ⎛ 1 ⎞
q1 = ⎜⎜ ⎟⎟ ,. ................................................... (1) RVIT ≈ ∆ .. ................................................... (8)
L1 ⎝ R v + R ∞b ⎠ πDio LkVIT
Now consider the analogous heat flow field around a simple
∆T ⎛ 1 ⎞ ,. ................................................... (2)
q2 = ⎜⎜ ⎟⎟ fin (Fig. 3). The heat flow pattern is two-dimensional as with
L2 ⎝ Rc + R∞c ⎠ the VIT heat flow pattern, and the resistance associated with
the fin can be calculated directly
q 3 = q 3 (h∞ (~
x ), ~
x , k st , ∆T ) ,. ........................................... (3) Temperature in Outer and Inner Pipes near 5.5" x
4.5" VIT Coupler, Heat Flux from OP, T1=150 F
where ~ x (x , r ) .
x =~ 250 160
150
In most cases, Rv >> R∞b and Rc << R∞c and the above 200 Temp OP/Coupler 140
Radial Heat Flux

Temperature [F]
equations become: Temp IP/Weld 130
[Btu/hr-ft2]

150 Heat Flux, Radial 120

110
q1 = ∆T ,.............................................................. (4) 100 100
L1 Rv
90
50 80
q 2 = ∆T .. ........................................................... (5) 70
L 2 R ∞c
0 60
Analysis of vacuum insulated tubing would be simple if 4 3 2 1 0
heat flow could be separated into vacuum and coupling Distance from Coupler Centerline [ft]
components and only Eqns. (4) and (5) were involved. Then,
total string heat loss would merely require a scaling of the heat Fig. 2 – VIT temperature profile near coupler. Heat flux is radial
2
from VIT exterior. T = 150 F, T∞ =80 F, h∞ =3 Btu/hr-ft -°F, k-value
loss of each component with the string length of each:
= 0.007 Btu/hr-ft-°F.

q1 L1 + q 2 L 2 = qT (L1 + L 2 ) ……………………………(6)
SPE 90151 3

environment. As a consequence, k -value (or VIT resistance)

is not a quantity which is intrinsic to the VIT, dependent only
T∞, h on VIT geometry, material composition and vacuum integrity.
Q , T1 It is a complex attribute requiring an analysis of the conditions
Perimeter P, surrounding the VIT.
Cross-Section S,
Thermal Cond. kF The vacuum space has aluminum foil wraps separated by
scrim, a non-conducting material. There is a metal getter
Length L which scavenges residual gas and preserves the long-term
vacuum viability. The thermal conductivity of the vacuum,
Fig. 3 – Simple fin of constant cross-sectional area.
k v , is a function of three components that act like resistors in
The fin has a simple analytical expression for steady-state parallel:
heat transfer, k v = k vrad + k vcond + k vconv ... ..................................... (16)
⎛ ⎞
Q FIN = hPk F S tanh ⎜ hP L ⎟(T1 − T∞ ) .. ....... (9) Because the vacuum is moderate (less than 100 millitorr)
⎝ k F S ⎠ the convective term can be neglected relative to the others.
The radiative term is nonlinear with respect to absolute
An analogous thermal resistance associated with the fin is temperature:
(
RFIN = T1 − T2
) = 1 ..(10) εσ ( Doi − Dio )(Tio2 + Toi2 )(Tio + Toi )
QFIN ⎛ ⎞ k vrad ≅ .......... (17)
hPk F S tanh⎜ hP L⎟ 4n
⎝ k F S ⎠
Thus k vrad is strictly a function of the reflective barrier
This resistance is a function of h , the external convection emissivity ( ε ) and the absolute temperatures of the two VIT
coefficient, and the thermal conductivity of the fin material. surfaces, namely the outer surface of the inner pipe and inner
The external convection coefficient can be related to the surface of the outer pipe, and n, the number of reflective
external resistance wraps, typically made of aluminum foil. The radiation
component is approximately 0.0003–0.0007 Btu/hr-ft-°F for
Rext , FIN ∝ 1 ,............................................................ (11) typical temperature ranges and number of foil layers.
h
The conductive term ( k vcond )is primarily a function of the
and R FIN can be related to RExt , FIN
residual gas conductivity. There is a negligible conduction
contribution from the minor contact between scrim and
R FIN ∝ 1
(
k F Rext tanh k F Rext ),.. .................... (12) aluminum foil. Conductivity of the residual gases in a VIT
equals approximately 0.001 Btu/hr-ft-°F, and is a weak
≈ kF
Rext
(
tanh k F Rext ) function of temperature. The vacuum k -value ( k v ) as
measured in the production facility varies between 0.0008 and
or, 0.003 Btu/hr-ft-°F, depending on a number of factors. It is
typically 0.0015 Btu/hr-ft-°F.
R FIN ≡ R FIN (Rext , k F ) ............................................... (13)
Finite Difference (FD) Model Development
VIT shares two characteristic with the fin: the heat flow A finite difference code was created to model VIT
pattern is two-dimensional axisymmetric and is a function of performance. This model is two-dimensional axisymmetric
the thermal conductivity of the material and the external with one-dimensional elements. A half section of the VIT is
thermal resistance. In contrast with the fin, however, VIT modeled. By symmetry, boundary conditions at each end are
performance as measured by k -value is also a function of the insulated. A diagram showing the code setup and resulting
vacuum k -value, equations can be seen in Fig. 4. An equation for weld
RVIT ≡ RVIT (Rext , k st , k v ) ......................................... (14)
elements was modified somewhat for elemental continuity.
The model has three zones in which h∞ is uniform - at the
The analogy is complete. From Eqn. 8 it follows that, coupler, weld region (2 feet in extent beyond weld), and body.
The finite difference code relies on an explicit numerical
kVIT ∝ 1
RVIT (Rext , k st , k v )
... .................................. (15) scheme in which suitable time steps are chosen for stability in
each region7. This code has been benchmarked below and
This expression for kVIT is complicated and can only be with other studies of an in-ground vertical experiment1.
solved numerically, in contrast with the fin example, which Standard k -value Definition
has a relatively simple analytical expression.
Historically, American engineers have used k -value as a
The most important observation from this discussion
regarding VIT is that, similarly to a fin, the resistance of the suitable metric for pipe-in-pipe insulation systems. The k -
VIT (or k -value) depends on the external resistance of the value concept is based on a one-dimensional analogy of heat
transfer from one fixed temperature to another,
4 SPE 90151

defined relative to the outer diameter of the outer pipe and

h∞, T∞ inner diameter of the inner pipe, and
Radiation Radial Conduction Outer Tube

q ln⎛⎜ oi ⎞
D
⎟
⎝ D io ⎠
k2 = ... .................................................. (21)
2π (T1 − T∞ )
Vacuum Space h1, T1 Axial Conduction Inner Tube defined relative to the inner diameter of the outer pipe and the
Element Discretization outer diameter of the inner pipe.
∆x NonDimensional k1/kv vs. NonDim Resistance for
Different Configurations, Values of kv.
• • • dr 1000
5.5x4.5 kv15
4.5x3.5 kv15
Ti+1 Ti Ti-1 7.0x5.5 kv15
3.5x2.3 kv15
Inner Pipe Equation (Pipe 1) 5.5x4.5 kv30
100 4.5x3.5 kv30
ρ1s A1x ∆x cp T1,i m+1– T1,im = - k1s A1x ( T1,i+1 m – 2 T1,im + T1,i-1 m ) 7.0x5.5 kv30

k1/kv
3.5x2.3 kv30
∆t ∆x
10
- kv A1 ( T1,i m – T2,im) / ∆ – h1 Ai1 (T1,im – T2,im )

Outer Pipe Equation (Pipe 2)

1
ρ2s A2x ∆x cp T2,i m+1– T2,im = - k2s A2x ( T2,i+1 m – 2 T2,im + T2,i-1 m )
0.00001 0.0001 0.001 0.01 0.1 1
∆t ∆x Rext / Rint
Fig. 5 – Nondimensional k-value (k1/kv) vs. Nondimensional
- h∞,im A2 (T2,im -T∞m) + kv A1 (T1,im – T2,im ) / ∆ External Resistance Rext/Rint = 2 kv / (ln(Doi/Dio) h∞ Doo) for
2
kv=0.0015, 0.0030 Btu/hr-ft-°F, h1=100 Btu/hr-ft -°F and
Fig. 4 - Finite difference code diagram showing configuration, configurations given.
discretization scheme, and equations for both inner and outer
pipe. NonDimensional k2/kv vs. Nondim Resistance for

q ln⎛⎜ 2 ⎞⎟
D Different Configurations, Values of kv.
100
⎝ D1 ⎠ 5.5x4.5 kv15
k − value ≡ .......................................... (18) 4.5x3.5 kv15
2π (T1 − T∞ ) 7.0x5.5 kv15
3.5x2.3 kv15
for a concentric cylindrical geometry, where k − value is 5.5x4.5 kv30
4.5x3.5 kv30
defined relative to D1 . 7.0x5.5 kv30
k2/kv

3.5x2.3 kv30
Alternately, other engineers traditionally use U -value, 10
defined as
2k − value ... .................................... (19)
U − value ≡
D1 ln ⎛⎜ 2 ⎞⎟
D
⎝ D1 ⎠
1
for a cylindrical geometry, where U 1 is relative to D1 .
0.00001 0.0001 0.001 0.01 0.1 1
The above k -value definition (and, by implication, the Rext / Rint
U -value) is strictly valid for one-dimensional radial heat Fig. 6 – Nondimensional k-value (k2/kv) vs. Nondimensional
transfer. Herein lies the subtlety, for the simple k -value External Resistance Rext/Rint = 2 kv / (ln(Doi/Dio) h∞ Doo) for
2
definition often misleadingly implies a simple interpretation of kv=0.0015, 0.0030 Btu/hr-ft-°F, h1=100 Btu/hr-ft -°F and
physical phenomena. configurations given.

For a consideration of a standard k -value we consider two

Simulations using the finite difference code were run for a
candidates commonly used:
number of standard VIT configurations and the results plotted
q ln⎛⎜ oo ⎞
D in Figs. 5 and 6. Nondimensionalized k -value is plotted with
⎝ Dii ⎟⎠ respect to nondimensionalized external resistance, defined by:
k1 = ...................................................... (20)
2π (T1 − T∞ ) Rext 2k v ..................................... (22)
=
h∞ Doo ln⎛⎜ 2 ⎞⎟
Rint D
⎝ D1 ⎠
SPE 90151 5

At first observation, the ranges of k1 are much higher than between horizontal annuli formed between concentric and
those of k 2 . At any one resistance level, the spread of k1 eccentric circular cylinders are well established5,6. These data
are here reexamined for possible correlations with overall VIT
values are at least two times higher than those of k 2 . The performance.
second observation is that the k 2 values converge to 1 as The methodology originally adopted for calculating U -
Rext Rint approaches unity while the k1 values converge to value2, based on the difference in heat loss from two different
sink temperatures, can be inaccurate for VIT. Calculations
a range of values that depend on the configuration. This has a
relied on the difference between two large numbers (heat loss
physical significance in that it means that as the external
at low sink temp – heat loss at high sink temp), each of which
resistance of the VIT increases, the insulation system
contained some experimental error, to eliminate the small but
performance as expressed by k 2 converges to the pure unknown end losses. This resulted in inconsistent values for
vacuum value. Thus, by choosing k 2 as a definition, it is easy U ii ; some were negative, a physical impossibility. An
to recognize how well the insulation system performs relative alternate method was used here, relying on averaged heat
to an ideal performance level. flows for two sink temperatures and assuming a small end loss
In typical Gulf of Mexico applications, 2 < k 2 / k v < 4, at the insulated VIT ends, calculated using first principles.
independent of configuration. As a practical measure, k 2 is a Results calculated in this manner are quite consistent.
significantly better choice to use as a definition. It is adopted Q = U ii Ai (T1 − T∞ ) + U a Aa (T1 − T∞ ) ......................... (24)
here and recommended for use,
where U a Aa (T1 − T∞ ) is the end loss term, which is
q ln⎛⎜ oi ⎞
D
⎝ Dio ⎟⎠ calculated for each configuration (these were 2-15% of Q ).
k − value ≡ k io = ................................... (23)
2π (T1 − T∞ ) There are two basic configurations in the Marlin experiments,
labeled here A and B. Configuration A has two annuli, 5-1/2
The choice of T∞ in the definition of Eqn. (23) should not in. x 10-¾ in. inner and 10-¾ in. x 16 in. outer. Configuration
be arbitrary. Problems arise in trying to choose a suitable T∞ B has one large annulus, 5-1/2 in. x 16 in. Configuration A
corresponds to Tests 1A-1D, 2A-2D, 7A-7G2. Configuration
in order to isolate the VIT from the environment so as to B corresponds to Tests 3A, 4A, 5A, and 6A (see Tables 1 and
describe VIT performance only, and not the performance of 2). Coolant mix was 35% glycol and 65% water. Glycol flow
the composite system. VIT performance cannot be stated in the outer annulus of Configuration B was approximately
independently of the environment. Therefore the choice of T∞ 900 - 950 gpm. Flow in Tests 1D, 2A, and 7A of
should be the undisturbed temperature away from temperature Configuration A, in which glycol flowed in both annuli, had
gradients caused by the presence of the VIT. One may use, inner annulus flow 250 – 350 gpm and outer annulus flow 600
for instance the sink temperature of a fluid flowing outside the gpm.
VIT as T∞ , as long as over its length, T∞ , the sink While there were subtle differences in how the VIT was
temperature, does not change more than an amount centered, whether temperature probes were used, whether the
comparable to the precision in either the measurements or the tubes were flooded or not, analytical work indicates that these
calculation. Such a change might occur if the Reynolds differences were not significant enough to perform special
number of the fluid is high enough, but these cases are rarely calculations. All calculations assumed the VIT was centered
encountered. For a riser application, T∞ should be the and the annuli were clear of obstructions. The three-
dimensional nature of the flow due to axial spreading is
undisturbed sea current temperature. In a downhole
another inaccuracy. However, because the ratio ( D 2 -
application, it should be the undisturbed ground temperature.
Because of limitations imposed by the definition of T∞ , VIT D1 )/ 2 Lhot of annular clearance to hot zone length is about 2
thermal performance as expressed by k -value is necessarily a in./48 in. = 4% for the natural convection cases of
Configuration A in the 5-1/2 in. x 10-¾ in. annulus,
composite performance, encompassing all the external
correlations are quite valid to this approximation level. For
resistances in the problem, including ground, annular, and
casing resistances. the correlation, k v = 0.005 Btu/hr-ft-°F was the best fit to
data.
Marlin Flow Loop Experiments The analysis relies on correlations in the literature for
Until recently, there was a lack of information regarding natural convection of horizontally aligned parallel tubes5,6.
fundamental mechanisms influencing VIT thermal Thermal properties of fluids are evaluated at the mean bulk
performance. While the Marlin flow loop tests2 suggested temperature in the annulus. Heat transfer rates of the steam
answers, important questions remained. Pipe orientation in condensing on the inside of the inner pipe range from 1100 –
these tests was horizontal, while the completion is vertical. 3700 Btu/hr-ft2, depending on the temperature at which the
Despite this drawback, the Marlin flow loop tests are the steam condenses, particularly on the inside on the coupler
most extensive series of lab controlled tests to date to simulate zone6.
downhole behavior. These tests serve a useful purpose in Calculations were performed using the finite difference
correlating boundary conditions to thermal performance, as code developed specifically for VIT evaluation. Fifty
empirical correlations for two-dimensional heat transfer elements were used for the 20 ft outer pipe, forty-nine for the
6 SPE 90151

20 ft inner pipe and four for the half coupler. The finite VIT generally performs well within a window of a few days
difference model accounts for increased thermal resistance due following startup. Its use is advantageous if shut in is highly
to coating and Teflon koozie insulation used for each case improbable. Clearly, the choice of whether to use VIT in
with such insulation, as well as the distance from weld to these situations is complex and depends on well parameters.
weld. Agreement between experimental and theoretical
methods is excellent (Fig. 7, R2=0.88).
The Marlin tests concluded that even moderate insulation Table 1 – Marlin Flow Loop Test Results and Finite
around the coupler region was very effective in decreasing Difference Model Calculations for Configuration A.
heat loss under adverse boundary conditions. Tables 1 and 2 Test A v e . W e ld A n n u l u s F lo w
t o W e ld [g p m ] M odel D a ta
give the data for U ii and k io and the calculated results of k io D is t a n c e
[ in ] /
C o u p lin g In n e r O u te r k io UA U ii k io
based on the given boundary conditions (flow rate and coolant In s u la t io n
1A -
properties, pipe geometry). Convection coefficient U ∞ based N itro g e n 5 .0
/N o n e
N a tu ra l
C o n v e c t io n
950 0 .0 0 4 9 1 .8 0 .2 1 3 3 0 .0 0 4 1

on either natural or forced convection is also given. 1B – 5 .0 N a tu ra l

950 0 .0 0 5 7 3 .3 0 .3 5 0 4 0 .0 0 6 8
F lu id 1 /N o n e C o n v e c t io n
1C –
5 .0 N a tu ra l
k-value, Theoretical vs. Experimental, BP 2% K C l
/N o n e C o n v e c t io n
950 0 .0 2 9 6 91 1 .1 5 8 1 0 .0 2 2 6
B rin e
Flow Loop Tests, 2000 1D –
5 .0
G ly c o l 320 630 0 .0 3 7 1 130 1 .9 8 3 4 0 .0 3 8 8
0.1 F lo w in g
/N o n e
2A
5 .0
G ly c o l 320 590 0 .0 3 7 1 130 2 .5 1 0 4 0 .0 4 9 1
R^2 = 0.88 F lo w in g
/N o n e
2B – 5 .0 N a tu ra l
N itro g e n 950 0 .0 0 4 9 1 .8 0 .2 9 9 0 0 .0 0 5 8
/N o n e C o n v e c t io n
k-value Theory

2C –
[Btu/hr-ft-F]

5 .0 N a tu ra l
F lu id 1 950 0 .0 0 5 7 3 .3 0 .3 3 9 1 0 .0 0 6 6
/N o n e C o n v e c t io n
2D – 5 .0 N a tu ra l
F lu id 2 950 0 .0 0 6 7 5 .8 0 .2 3 2 0 0 .0 0 4 5
/N o n e C o n v e c t io n
0.01 7A –
8 .6 /C o a te d
G ly c o l 310 910 0 .0 0 9 7 120 0 .4 7 4 5 0 .0 0 9 2
+ K o o z ie
F lo w in g
7B –
8 .6 /C o a te d N a tu ra l
G ly c o l 980 0 .0 0 8 8 42 0 .4 4 1 3 0 .0 0 8 6
+ K o o z ie C o n v e c t io n
S ta tic
7C –
1 G ly c o l
8 .6 /C o a te d N a tu ra l
1000 0 .0 0 8 8 42 0 .4 1 8 6 0 .0 0 8 1
+ K o o z ie C o n v e c t io n
S ta tic
1 7D –
N itro g e n
8 .6 /C o a te d N a tu ra l
950 0 .0 0 4 9 1 .8 0 .2 3 9 1 0 .0 0 4 6
+ K o o z ie C o n v e c t io n
0.001 7E – 8 .6 /C o a te d N a tu ra l
F lu id 1 920 0 .0 0 6 1 3 .3 0 .2 7 2 0 0 .0 0 5 3
+ K o o z ie C o n v e c t io n
0.001 0.010 0.100 7F –
k-value Experiment [Btu/hr-ft-F] F lu id 1 ,
D is a b le d
8 .6 /C o a te d
+ K o o z ie
N a tu ra l
C o n v e c t io n
920 0 .0 4 0 4 3 .3 2 .5 7 7 5 0 .0 5 0 4
V IT
Fig. 7 - Updated Marlin Flow Loop Test results compared with 7G –
N 2, 8 .6 /C o a te d N a tu ra l
1000 0 .0 2 3 1 1 .8 1 .3 8 1 1 0 .0 2 7 0
finite difference model calculations. Best fit is k v =0.005 Btu/hr-ft- D is a b le d + K o o z ie C o n v e c t io n
V IT
°F.
2
U n its o f k a r e B t u / h r - f t - ° F , u n its o f U a r e B t u / h r - f t -° F

Transient Effects of VIT

The time constant ( τ ) of VIT is measured in hours, and is
highly dependent on boundary conditions. For instance, in a
typical 5-1/2 in. x 4-1/2 in. VIT with no coupling insulation in Table 2 - Marlin Flow Loop Test Results and Finite
the Gulf of Mexico environment, with h∞ = 10 Btu/hr-ft-°F, Difference Model Calculations for Configuration B
Ave. Weld to Weld Annulus
τ ~ 5 hours. With coupling insulation, and h∞ = 5 Btu/hr-ft- Test Distance [in]/ Coupling Flow UB
kio
Model
Uii
Data
kio
Data
Insulation [gpm]
°F, τ ~ 10 hours1. In a vertical convective environment, roll 3A
2.5
950 100 0.0231 1.0913 0.0213
/None
cells form in brine-filled annuli whose velocities are in the 5- 4A 8.63/Coated + Koozie 800 89 0.0093 0.4545 0.0089
10 cm/s range. These cells form rapidly, and generally take 5A 8.75/Koozie, No Coating 980 100 0.0112 0.5020 0.0098
less than one hour to reach a periodic steady state. 6A 8.63/Coating, No Koozie 970 100 0.0188 0.7657 0.0149
Units of k are Btu/hr-ft-°F, units of U are Btu/hr-ft2-°F
Time constants are apparent from a simulation of shut in
following startup operations3. Fig. 8 shows typical wellhead
temperature curves following well start up for shut in 12 and
48 hours later, comparing VIT and bare pipe. Note that there Not only is VIT k -value important in a consideration of
is a crossover point beyond which bare pipe is more time constants; external variables such as ground conductivity,
advantageous to use. In the first few hours of using VIT, density and specific heat also play a role. The time constant
wellhead temperatures climb above the hydrate point (usually for ground soak can exceed two weeks to get within several
65 – 90 °F). Bare pipe takes longer to rise above a given degrees of final value. Annular temperatures are affected by
hydrate temperature, sometimes several days. If the crossover this, but on a time scale much longer than that of VIT.
point is higher than the hydrate temperature, however, then Thus there are only a small number of time constants to
following shut in and beyond the crossover point, bare pipe consider. Convection cells in the outer annuli require less than
takes longer to reach the hydrate point. This is explained by one hour to develop, the VIT requires on the order of several
ground soak. Paradoxically, bare pipe accelerates ground hours to react to changes in temperature, and the ground
soak, providing a “warm blanket” effect following shut in. requires several weeks to reach a quasi-steady state. Annular
SPE 90151 7

natural convection does not migrate upwards indefinitely; especially effective when placed over the weld. As a rough
there is a two-dimensional balance in radial and axial heat illustration, compare a 5-1/2 in. (4.892 in. ID) x 4-1/2 in. VIT
flow that leads to stable thermal distributions. This can be with uninsulated coupler with h∞ = 5 Btu/hr-ft2-°F, k v =
demonstrated by an examination of the Marlin fiber optic
0.0015 Btu/hr-ft-°F, and k io = 0.0045 Btu/hr-ft-°F. With a ¼
temperature data4, where temperature patterns have been
stable for several years. in. polyurethane wrap, using a resistance in series calculation,
h∞ ~ (1/5 Btu/hr-ft2-°F + 0.25 in./(12 in./ft. 0.08 Btu/hr-ft2-
Shut In Curves Following Startup, T1=170 F,
°F))-1 = 2.1 Btu/hr-ft2-°F. Thus through the use of a ¼ in.
T inf = 70 F.
180 polyurethane sleeve, h∞ has dropped by over a factor of two.
VIT
Shut In, From Fig. 7, 2 k v /(ln( Doi / Dio ) h∞ Doo ) = 2 (0.0015)
160
Shut In, 12 hours 48 hours
Wellhead Temp [F]

/(ln(4.892/4.5) (2.1) 5.5/12) = 0.037, then k io / k v = 2

140
approximately, or k io = 0.003 Btu/hr-ft-°F. Therefore,
120
through the use of a polyurethane sleeve, k -value has been
Crossover reduced by approximately 33%.
100
Points
80 Conclusions
Bare Pipe A discussion of the physics of VIT thermal performance under
60 widely varying conditions has been presented. Models are
0 20 40 60 80 developed that agree with data to approximation levels of 10-
Time [hr] 15%. Thermal performance enhancement generally leads to a
Fig. 8 – Shut in curves for VIT and bare pipe following flow choice of appropriate insulation around the coupler and weld
startup. Results compare relative advantages of each. regions.
Several factors have led to the development of a useful
Discussion theory of VIT thermal design. These include the accessibility
The Marlin flow loop tests provided an outstanding of computer-based modeling approaches, a small number of
opportunity to benchmark computational models. As is relevant field test data, correlations of thermal phenomena in
demonstrated in the present study, the thermal effects of VIT the literature, and analytical models from theoretical
can be accurately modeled and the sensitivity of k -value to discussions in textbooks. By far the most useful single study
boundary conditions is apparent. Typical ranges of external was the Marlin flow loop tests. From this study, it is possible
boundary conditions ( h∞ in the hot coupler zone) for 4-1/2 in. to correlate data with theoretical results using computer-based
x 3-1/2 in. standard VIT, k v =0.0015 Btu/hr-ft-°F, h1 =100 modeling together with established correlations of known
forced and free convection phenomena. Resulting agreement
Btu/hr-ft2-°F, without coupling insulation are as follows: between theory and data is excellent. Larger implications for
• Gulf of Mexico completion (T1 < 200°F): 5 < h∞ < thermal performance under widely varying applications are
15 Btu/hr-ft2-°F, 0.0053 < k io < 0.010 Btu/hr-ft-°F, clear. Results of this study lead to a more rational thermal
design methodology.
• Flowing gas lift in annulus outside VIT: 10 < h∞ <
30 Btu/hr-ft2-°F, 0.008 < k io < 0.016 Btu/hr-ft-°F, Nomenclature
A Outer surface area of element, [L2], ft2
• Riser, two annuli filled with gelled brine: 2 < h∞ <
Aa Inner surface area of inner pipe at pipe ends, [L2], ft2
5 Btu/hr-ft2-°F, 0.0036 < k io < 0.0053 Btu/hr-ft-°F, Ai Inner surface area of inner pipe, [L2], ft2
• Riser, two annuli filled with brine: 30 < h∞ < 60 Ax Cross sectional area of element, [L2], ft2
2
Btu/hr-ft -°F, 0.016 < k io < 0.023 Btu/hr-ft-°F, Specific heat of steel, [L2/(T2Θ)], Btu/hr-lb-°F
cp
• 8
Steamflood, refluxing annulus , steam at 550°F, dr Thickness of element, [L], ft
h1 =1000 Btu/hr-ft2-°F: 4 < h∞ < 10 Btu/hr-ft2-°F, Dii Inner diameter of inner pipe, [L], ft
0.0050 < k io < 0.0091 Btu/hr-ft-°F. Dio Outer diameter of inner pipe, [L], ft
The effect of ground conductivity on k -value is very Doi Inner diameter of outer pipe, [L], ft
small. A factor of 2 decrease in k ∞ results in a decrease in k - Doo Outer diameter of outer pipe, [L], ft
value of approximately 4%. D1 Diameter corresponding to T1 , [L], ft
A precise calculation of h∞ must be done iteratively. D2 Diameter corresponding to T2 , [L], ft
Current practice uses either ¼ in. thick Teflon internal
h Heat transfer coefficient associated with fin at T∞ ,
coupling insulators (koozies) or ¼ in. – ¾ in. thick external
polyurethane wraps. Both are significantly effective at [ML/(T3Θ)], Btu/hr-ft2-°F
reducing effective h∞ and thus k io and the external wrap is
8 SPE 90151

h1 Internal heat transfer coefficient associated with T1 , R FIN Thermal resistance of fin, [ΘT3/(ML2)], hr-°F/Btu
[ML/(T3Θ)], Btu/hr-ft2-°F Rv Thermal resistance of vacuum chamber,
h∞ Heat transfer coefficient associated with T∞ , [ΘT3/(ML2)], hr-°F/Btu
[ML/(T Θ)], Btu/hr-ft2-°F
3 RVIT Thermal resistance of VIT, [ΘT3/(ML2)], hr-°F/Btu
kF Thermal conductivity of fin, [ML/(T3Θ)], Btu/hr-ft- R ∞b Thermal resistance outside VIT body area,
°F [ΘT3/(ML2)], hr-°F/Btu
k io Thermal conductivity associated with outer diameter R ∞c Thermal resistance outside VIT coupler area,
of inner pipe, [ML/(T3Θ)], Btu/hr-ft-°F [ΘT3/(ML2)], hr-°F/Btu
k st Thermal conductivity of steel, [ML/(T3Θ)], Btu/hr-ft- S Cross-sectional area of fin, [L2], ft2
°F Tio Temperature at inner pipe, outer diameter, [Θ], °R
kv Thermal conductivity of vacuum, [ML/(T3Θ)], Toi Temperature at outer pipe, inner diameter, [Θ], °R
Btu/hr-ft-°F T1 Internal temperature, [Θ], °F
k vrad Thermal conductivity of vacuum chamber due to T∞ External environmental temperature, [Θ], °F
radiation, [ML/(T3Θ)], Btu/hr-ft-°F Ua Overall heat transfer coefficient of VIT ends relative
k vcond Thermal conductivity of vacuum chamber due to
to inner diameter of inner pipe, [M/(T3Θ)], Btu/hr-ft2-
conduction, [ML/(T3Θ)], Btu/hr-ft-°F °F
k vconv Thermal conductivity of vacuum due to convection, U ii Overall heat transfer coefficient relative to inner
[ML/(T3Θ)], Btu/hr-ft-°F diameter of inner pipe, [M/(T3Θ)], Btu/hr-ft2-°F
kVIT Thermal conductivity of VIT, [ML/(T3Θ)], Btu/hr-ft- U∞ Overall heat transfer coefficient associated with T∞ ,
°F [M/(T3Θ)], Btu/hr-ft2-°F
k1 Thermal conductivity, definition 1, [ML/(T3Θ)], ∆ Vacuum gap width, [L], ft
Btu/hr-ft-°F x Axial coordinate, [L], ft
k2 Thermal conductivity, definition 2, [ML/(T3Θ)], ~
x Distance vector, [L], ft
Btu/hr-ft-°F ε Emissivity of vacuum insulation wrap
L Fin length or length, [L], ft σ Stephan-Boltzmann constant, [M/(T3Θ4)], 1.71 x 10-9
L1 VIT body length, [L], ft Btu/hr-ft2-°R4
L2 Coupler length, [L], ft
References
m Time index 1. Azzola, J.H., et al: “Application of Vacuum Insulated
n Number of reflective wraps of radiation filler Tubing to Mitigate Annular Pressure Buildup,” paper SPE
P Perimeter of fin, [L], ft 90232 presented at the 2004 Annual Technical Conference
q Heat flow loss per unit length, [ML/T3], Btu/hr-ft and Exhibition, Houston, Texas, 26-29 September.
2. Ellis, R.C., et al: “Marlin Failure Analysis and Redesign;
qT Heat flow loss per unit length across VIT joint,
Part 2, Redesign,” paper SPE 74529 presented at the 2002
[ML/T3], Btu/hr-ft IADC/SPE Drilling Conference, Dallas, Texas, 26-28
q1 Heat flow loss per unit length across the vacuum February.
chamber, [ML/T3], Btu/hr-ft 3. Feeney, S.: “Vacuum-Insulated Tubing: Project Case
Histories, Future Applications,” Offshore, (October 1999)
q2 Heat flow loss per unit length across the coupling, pp. 100, 169.
[ML/T3], Btu/hr-ft 4. Gosch, S.W., et al: “Marlin Failure Analysis and Redesign;
q3 Heat flow loss per unit length in the weld region, Part 3, VIT Completion With Real-Time Monitoring,”
paper SPE 74530 presented at the 2002 IADC/SPE Drilling
[ML/T3], Btu/hr-ft Conference, Dallas, Texas, 26-28 February.
Q Heat Loss, [ML2/T3], Btu/hr. 5. Kakac, S., Shah, R.K., and Aung, W.: Handbook of Single-
QFIN Heat flow loss from fin, [ML2/T3], Btu/hr Phase Convective Heat Transfer, John Wiley & Sons
(1987).
r Radial coordinate, [L], ft 6. Bejan, A.: Convection Heat Transfer, John Wiley & Sons,
rio Inner pipe outer radius, [L], ft 2nd Edition (1995).
R Thermal resistance, [ΘT3/(ML2)], hr-ft-°F/Btu 7. Holman, J.P., Heat Transfer, McGraw Hill, 5th Edition,
Rc Thermal resistance of coupler, [ΘT3/(ML2)], hr- 1981, pp. 135-155.
8. Willhite, G.P., Griston, S.: “Wellbore Refluxing in Steam
°F/Btu Injection Wells,” paper SPE 15056 presented at the 1986
Rext External thermal resistance, [ΘT3/(ML2)], hr-°F/Btu California Regional Meeting, Oakland, California, 2-4
April.
Rext , FIN External thermal resistance for fin, [ΘT3/(ML2)], hr-
°F/Btu
Rint Internal thermal resistance, [ΘT3/(ML2)], hr-°F/Btu