OTC 17894
Flow-Assurance Modeling: Reality Check and Aspects of Transient Operations of
Gas/Condensate Pipelines
L. Hagesæther, K. Lunde, and F. Nygård, Statoil, and H. Eidsmoen, Scandpower Petroleum Technology
Copyright 2006, Offshore Technology Conference
This paper was prepared for presentation at the 2006 Offshore Technology Conference held in
Houston, Texas, U.S.A., 1–4 May 2006.
This paper was selected for presentation by an OTC Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Offshore Technology Conference and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Offshore Technology Conference, its officers, or members. Papers presented at
OTC are subject to publication review by Sponsor Society Committees of the Offshore
Technology Conference. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes without the written consent of the Offshore Technology
Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not
more than 300 words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, OTC, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
Abstract
Statoil operates a number of gas condensate pipelines in the
North Sea. This paper focuses on experience gained from
operation and simulation, up to and including tail end, of the
150 km long 22” Huldra to Heimdal pipeline.
During initial production at Huldra the liquid accumulation
was higher than predicted by modeling. Additionally liquid
surge waves not found in simulations were observed at the
receiving facility. These findings were challenging due to very
low liquid surge capacity in the receiving facility, and the
minimum flow rate was increased. Investigations were carried
out in order to explain the observations, and it was determined
that condensate carry over in the Huldra scrubber significantly
influenced the condensate content in pipeline. The pipeline has
now entered the tail end production phase. Due to high liquid
content at low production rates, water/MEG no longer reaches
the receiving facility on a regular basis, causing local hydrate
problems at Heimdal and a lack of MEG for re-injection.
Simulations show that cyclic operation of the receiving facility
will transport water/MEG out of the pipeline on a more
regular basis. This change in operating philosophy may be
implemented in 2006.
Introduction
The traditional gas condensate development had a process
facility close to the field, and the fluids would be exported
through a single phase gas line and a single phase liquid line.
This involved expensive process facilities at remote locations,
often off-shore, and multiple pipelines. From the 1970’s
significant research efforts was put into multiphase transport.
As the ability to design two phase (gas and hydrocarbon
liquid) pipelines advanced it was sufficient to dehydrate the
fluid close to the field, and the hydrocarbons were transported
in a multiphase pipeline. This gave significant saving due to a
simpler and smaller process facility and a single pipeline
(MLNG [1] in Malaysia, BLNG in Brunei, Sable Island in
Canada and Nam Con Son in Vietnam). The last step in the
development has been the ability to design three phase
pipelines (gas, hydrocarbon liquid and water), which was
achieved in the early 1990’s. This completely removes the
need for the process facility, which are replaced by simple
wellhead platforms (Huldra [2,3] and Troll [4] in Norway,
South Pars in Iran, Ras Laffan and Qatar Gas in Qatar and
Goldeneye in the UK) or complete sub-sea developments
(Mensa [5] and Canyon Express [6,7] in the US,
Scarab/Saffron [8] in Egypt, TOGI [9], Snøhvit and Ormen
Lange [10] in Norway). Table 1 and Figure 1 summarizes the
development of gas condensate pipelines by some of their key
parameters, such as pipeline diameter, length, two phase
versus three phase and development by wellhead platform
versus subsea development. The present trend is that almost
all pipelines are three phase, and most are subsea
developments unless the water is shallow.
The main issues considered when designing gas
condensate systems are usually pressure drop, liquid handling
and hydrate prevention. Pipeline pressure drop is mainly
related to selection of correct pipeline size, while liquid
handling relates to slug catcher size and plant liquid
processing capacity. A large diameter pipeline will usually
give a low pressure drop, but a high liquid content, causing
liquid handling problems, while a smaller pipeline diameter
will give higher pressure drop, but less liquid content. In
addition liquid handling and hydrate prevention are closely
tied to the operational procedures of the pipeline, for
operations such as rate changes, shut-in and start-up, blow-
down and pigging. Other potential issues considered are
corrosion, wax deposition and erosion.
Even with proper pipeline modeling during the design
phase there are still uncertainties in the simulations results.
This paper summarizes some of the experiences from
modeling and operation of gas condensate pipelines. The
simulation results are based on analysis using the multiphase
pipeline simulation tool OLGA®.
2 OTC 17894

The total flow rate in the two pipelines varies in the range
from 40 to 100 MSm3/d on a daily basis, depending on
delivery requirements. The last 10 km of the pipeline is very

400 km
steep, with 1500 m angled at 8°. In periods of low rate, below
approximately 30 MSm3/d in each pipeline, significant
amounts of condensate and water/glycol accumulates in the
pipeline, due to the steep angles, causing severe liquid surges
when the production rate is increased [4]. This explains the
100 km
need for a large onshore slug catcher.

Kvitebjørn was discovered in 1994 and production was

initiated from a wellhead platform in 2004. It is located in the
1980 2010
Norwegian part of the North Sea North-west of Bergen,
somewhat to the North-west of the Troll field. The recoverable
ra n s
le
za

at nø n

en Ga t

e
a

S on
S a
at ll A
I
G

La II
M as
n

ca Co res

i
o
G

ng
r

hv
s
ab
go

LN

lu

am x d

fr
en

S
O

s
G

reserves are approximately 52x109 Sm3 and the maximum

N n E ul
ro

af
er

p
T
or

ar

H
T
M

S
G

rm r
b

O a
production rate is 20.7 MSm3/d. The rich gas is exported to
Q

yo

Q
an

S
C

shore through a 150 km long 30” pipeline with a design

Figure 1. Summary of some of the main gas condensate capacity of 26.5 MSm3/d, which is parallel to the Troll
developments worldwide since 1980. The width and height of the pipeline the last 65 km towards shore. This pipeline utilizes
bars indicate diameter and length of the pipelines, two phase the same slugcatcher at Kolsnes as the Troll pipelines.
pipelines are shown in white and black while three phase lines are
shown in white, black and blue. Stabilized condensate is exported through a separate
condensate pipeline. Some of the considerations around
possible development options in combination with the Troll
Table 1. Summary of key information for some of the main gas pipelines have been presented previously [11]. There is
condensate pipelines worldwide. considerably less condensate in this pipeline than in the Troll
Field Length Diam Depth Subsea Prod. Fluid pipelines since condensate is separated and removed at the
[km] [in] [m] Dev. Start Kvitebjørn platform. Additionally gas production from the
Ormen Lange 120 2x30 1100 Yes 2007 3P Visund platform is tied-in through a 35 km long 20” pipeline 2
Qatar Gas II 90 32 50 No 2006 3P km downstream of Kvitebjørn.
Snøhvit 143 28 345 Yes 2005 3P
Goldeneye 105 20 120 No 2004 3P Mikkel and Midgard were discovered in 1981 and 1987
Kvitebjørn 150 30 190 No 2004 3P respectively and are both located at Haltenbanken. Mikkel is
Scarrab/Saffron 90 20 to 36 90 Yes 2003 3P located approximately 40 km south of Midgard and has
South Pars 2+3 105 2x32 65 No 2002 3P recoverable reserves of 24x109 Sm3 of gas, 5.9x106 ton of
Nam Con Son 399 24 180 No 2002 2P NGL and 6.7x106 Sm3 of condensate. Midgard is part of the
Canyon Express 92 2x12 2200 Yes 2002 3P Åsgard development which has total recoverable reserves of
Huldra 145 22 120 No 2001 3P 70x106 Sm3 oil, 193x109 Sm3 of gas, 35x106 ton of NGL and
Sable Island 225 26 No 1999 2P 46x106 Sm3 of condensate. Initially Midgard was developed
Sable Island 175 <20 No 1999 3P by a 35 km long 20” pipeline transporting the untreated well
Ras Gas 92 32 50 No 1999 2P fluids to the Åsgard FPSO. Subsequently Mikkel was tied in to
Mensa 101 12 1600 Yes 1997 3P the Midgard development by way of a single 37 km long 18”
Qatar Gas 82 32 50 No 1996 2P flowline.
Troll 65 2x36 300 No 1995 3P
Merluza 210 16 130 No 1993 2P Snøhvit was discovered in 1984 and production is
TOGI 48 20 300 Yes 1991 3P expected to start mid 2007. The development consists of the
MLNG 120 32 100 No 1983 2P three discoveries, Snøhvit, Askeladd and Albatross, and it is
located in the Barents Sea approximately 140 km Northwest of
Hammerfest. The recoverable reserves are 161x109 Sm3 gas
Statoil’s Gas Condensate Pipelines and 18x106 Sm3 of condensate and the production rate is 21
MSm3/d. The field is developed through a complete subsea
The Troll field was discovered in 1979 and production was solution, which feeds a LNG plant. Gas and condensate is
initiated in May 1996. It is located in the Norwegian part of transported to shore through dual 143 km long 28” pipelines
the North Sea, North-west of Bergen. Troll is the largest with a 3000 m3 onshore slugcatcher. The design capacity of
producing offshore gas field in Europe, with estimated the slugcatcher is 1900 m3 of condensate and 680 m3 of
recoverable reserves of 1325x109 Sm3 of gas and with a water/MEG. Since peak condensate rate and peak water rate
capacity of 100 MSm3/d. The field development consists of a will most likely not occur at the same time, the condensate
wellhead platform, two 67 km long 36” multiphase pipelines capacity is potentially larger than 1900 m3. The Snøhvit
for transfer of gas, condensate and water/MEG to shore and an development is probably the most complex gas condensate
onshore process facility with a 2500 m3 slugcatcher, of which development so far, due to the long pipelines and the very
2000 m3 is for condensate and 500 m3 is for water/MEG. harsh ambient conditions.
OTC 17894 3

All numbers for recoverable reserves and production rates
are taken from “Fact Sheet 2005 Norwegian Petroleum
Activity” [12].
The Huldra to Heimdal pipeline system
Huldra is a gas-condensate field located in the Norwegian
sector of the North Sea. The field was discovered 1982 and
production was started in 2001 from an unmanned wellhead
platform. The total recoverable reserves are 12.9x109 Sm3 of
gas and 4.7x106 Sm3 of oil. The produced rich gas is
transported as a multiphase mixture of gas, condensate and
aqueous glycol through a 22” diameter and 150 km long
pipeline to the Heimdal platform for final processing and
export. The pipeline goes through mildly undulating terrain at
depths between 95 and 125 m. The not stabilized condensate is
exported to Veslefrikk through a 16 km long 8” pipeline. The
overall field layout is shown in Figure 2 and the pipeline
profile from Huldra to Heimdal is shown in Figure 3.
Figure 5 shows the daily production rate from field start-up
through December 2005, while Figure 6 shows the thirty day
running average of the production rate. The production rate is
expected to continue declining through 2006. The minimum
flow rate was initially 6 MSm3/d, but this was raised to 8
MSm3/d when the water/MEG phase did not arrive as
expected at Heimdal after start-up. The rate dropped below 8
MSm3/d at the beginning of 2005 and below 6 MSm3/d at the
end of 2005.
An interesting observation during tail-end production is
that in the initial phase after a long shut-in the production rate
is higher than before the shut-in. This is due the reservoir
pressure equalizing during the shut-in, giving a relatively high
pressure close to the well, and a high production rate
immediately after restart.

The design rate for the rich gas pipeline is 10.35 MSm3/d
and the maximum flow rate is 12 MSm3/d. During the design
phase the lower limit for the gas flow rate was set to 6
MSm3/d. This rate was initially used at production start-up.
Both condensate and aqueous phase arrived later than
simulated, and the aqueous phase arrived after the rate was
increased above 6 MSm3/d. It was then decided to increase the
lower limit for the gas flow to 8 MSm3/d.

The pipeline outlet pressure at Heimdal is 100 bara, and

since Heimdal processes fluid from a number of fields this
pressure can not easily be changed. Heimdal has a 7 m3 inlet
separator (free water knock-out drum), a water/MEG capacity
of about 1 m3/h and a condensate capacity of approximately
60 m3/h available for the production from Huldra. Considering
the size and length of the pipeline the capacity of the inlet
separator is very limited, compared for instance to the Troll
development which has a 2500 m3 slug catcher. The main Figure 2. Schematic drawing of the Huldra to Heimdal pipeline
reasons for this is that the flow rate in the Huldra pipeline can system.
be kept relatively constant, avoiding major liquid surges,
increased confidence in the design tools and the requirement 40

to only make limited investments to make the development 20

economically feasible. 0

-20
Figure 4 shows a sketch of the process facilities at Huldra.
Depth [m]

-40
The production separator is used to separate liquid from gas.
The gas is then cooled in the heat exchanger before entering a -60

gas scrubber. MEG is added to the gas from the scrubber -80

before it is sent to Heimdal. Note that the platform uses a -100

floating pressure scheme, which means that the pressure in the
-120
gas scrubber and in the process separator are both dependent
-140
on the pipeline outlet pressure at Heimdal and the pressure 0 20 40 60 80 100 120 140 160
drop in the pipeline. There is typically a 2 bar pressure drop Distance [km]

between the separator and the scrubber, and between the Figure 3. Pipeline profile from Huldra to Heimdal.
scrubber and the pipeline inlet. The temperature is typically
100°C in the separator and 29°C in the scrubber.
4 OTC 17894

Glycol
(pH stab MEG)
of the pipeline, with maximum error of approximately 10% in
Gas cooler for flow rates in the range from 6.5 to 11 MSm3/d [2,3].
Gas Export

Allocation 31 136
Gas Scrubber Gas Metering Temperature
From wells Pressure
28 132

25 128

22 124
Condensate/

Temperature [C]
Production Water Export
Separator 19 120

16 116

13 112
Gas
Condensate 10 108
Water
Figure 4. Schematic of the process facilities at Huldra. 7 104

4 100
0 20 40 60 80 100 120 140 160
12 Distance [km]

3
Figure 7. Temperature and pressure profiles for 11 MSm /d steady
10 state.
0.06
Hold-up total liquid
8
Hold-up water
[MSm3/d]

0.05
6

0.04
4
[-]

2 0.03

0 0.02
0 250 500 750 1000 1250 1500
Time [days]
0.01
Figure 5. Measured mean daily flow rate from field start-up
through December 2005.
0
0 20 40 60 80 100 120 140 160
12 Distance [km]

3
Figure 8. Total liquid and water hold-up profiles for 11 MSm /d
10 steady state.

8
Figure 9 shows the pipeline pressure drop, total liquid
[MSm3/d]

content and water/MEG content as function of flow rates for

rates up to 7 MSm3/d, assuming no liquid carry over in the
6

4
scrubber and 25°C pipeline inlet temperature. This is
considered the most likely assumption for the late production
2 phase. From the curves we notice that the flow becomes
gravity dominated for flow rates below approximately 4
0 MSm3/d, when the pressure drop starts increasing with
0 250 500 750 1000 1250 1500
reduced flow rate. From the curves for liquid content we see
Time [days]
that especially the water/MEG phase accumulates at rates
Figure 6. Thirty day running average of daily flow rates from field below 5 MSm3/d. Due to the low water/MEG rate and the very
start-up through December 2005.
long accumulation time it is not likely that the pipeline will
ever operate at steady state for rates below 5 MSm3/d. If the
Pressure drop and liquid hold-up
pipeline was operated at a low rate for an extended period of
time the condensate would first build up and then it would
The steady state pipeline pressure drop for the Huldra to
slowly be displaced by the water/MEG phase.
Heimdal pipeline is in the range from 30 bar at the maximum
rate to a minimum of approximately 8 bar for 4 MSm3/d.
After start-up it was found that the liquid production from
Figure 7 and Figure 8 show temperature, pressure and hold-up
the pipeline was significantly higher than anticipated, which
profiles for 11 MSm3/d steady state with 60% condensate
meant that the liquid hold-up predictions were not accurate.
carry over in the Huldra scrubber. The measured pressure drop
This was mainly caused by condensate carry over in the
matches very well with the predictions done during the design
Huldra scrubber. Since the temperature drop over the heat
OTC 17894 5

exchanger is in the order of 70°C, condensation over the heat 1400

Total liquid content
35

exchanger is significant, and with a scrubber with low Water/MEG content

efficiency the liquid carry over can be larger than the 1200 30

condensation in the pipeline.

1000 25

Total liquid [m3]

Water/MEG [m3]
Figure 10 shows the total liquid and the water/MEG 800 20
content as function of the condensate carry over in the
scrubber for a flow rate of 11 MSm3/d, based on simulations. 600 15

As we can see from the curve the total liquid content has an
400 10
almost linear increase with the carry over, while the
water/MEG content is almost constant. It has been estimated 200 5
that the condensate carry over is 20% at 6 MSm3/d, 40% at 8
MSm3/d and 60% at 11 MSm3/d. It should be noted that for 0 0
0 10 20 30 40 50 60
gas condensate pipeline proper modeling of the pipeline is
Condensate carry over [%]
very important in order to obtain correct estimates of the
liquid content [13]. Figure 10. Total liquid and water/MEG content as function of
3
condensate carry over at 11 MSm /d.

It is assumed that there is no carry over of the water in the

scrubber, which has been verified by the rates measured at the 33.5
Pressure drop
pipeline outlet. Note that the amount of water/MEG injected 33
into the pipeline is larger than the amount of water condensed
in the pipeline. 32.5

32
While the liquid content in the pipeline is very sensitive to [bar]

the condensate carry over the pressure is not. Figure 11 shows 31.5

the pressure drop at 11 MSm3/d as function of condensate

31
carry over, and the difference between 0 and 60% carry over is
only 3 bar, which is less than 10%. 30.5

30
An additional complexity is that the pressure in the Huldra 0 10 20 30 40 50 60
process facility floats on the pipeline inlet pressure, which Condensate carry over [%]

means that gas composition will vary with the flow rate. Figure 11. Pressure drop as function of condensate carry over at
3
11 MSm /d.
Based on the above discussion it is clear that it is very
important to have correct input to a simulation model, when Liquid content measurements
designing the system and when later matching the simulation
model to field data. Liquid content, and particularly water content, is very difficult
to measure. It can be determined by pigging or rate changes,
4000
Water content
16 by analyzing slug catcher levels and drain rates, and by tracer
3500
Total liquid content
14
technology [14,15]. For the Troll pipeline the liquid content
Pressure drop has been estimated by pigging [4], but this is not an option for
3000 12 the Huldra pipeline due to limitations of the receiving facility
at Heimdal.
2500 10

The basic idea behind measurement by tracer is adding a

[bar]
[m3]

2000 8

tracer at the pipeline inlet and then measure the mean transport
1500 6
time through the pipeline. This is achieved by continuously
1000 4 measuring the tracer concentration at the pipeline outlet and
by determining the time from the tracer is first injected at the
500 2
pipeline inlet until it arrives at the pipeline outlet. By adding
0 0
one tracer that goes in the condensate phase and another that
2 3 4
[MSm3/d]
5 6 7
goes in the aqueous phase both the condensate and
MEG/water content in the pipeline can be estimated. In theory
Figure 9. Steady state pressure drop, total liquid hold-up content this is simple, but there are significant technical challenges in
and water content as function of flow rate, assuming no
condensate carry over in the Huldra scrubber. selecting and handling the tracers and performing accurate
measurements.

For a pipeline with considerable condensation the liquid

rate will increase gradually along the pipeline. Since the tracer
in general will partition to the liquid phase the condensation
6 OTC 17894
will influence the residence time and an initial estimate of the
liquid hold-up along the pipeline is required for the
computation of the pipeline liquid content. That is, this ends
up being an iterative process where the hold-up estimates can
gradually be improved.
An additional complication for the Huldra to Heimdal
pipeline is that there is no accurate measurement of the
condensate and water rates arriving at Heimdal, the
condensate rate can only be estimated from the average
condensate export rate. As discussed previously there is
considerable condensate carry over in the scrubber at Huldra,
the condensate rate can not be estimated from the
condensation in the pipeline alone. However, by determining
the concentration of the tracer in the inlet flow at Heimdal
during steady state these quantities can be estimated.
Institutt for Energiteknikk (Institute for Energy
Technology, Kjeller, Norway) has developed technology to
perform such tracer measurements as described above and
performed tracer measurement for two steady state flow rates
in the Huldra to Heimdal pipeline in 2003. The rates chosen
were 8 MSm3/d and 11 MSm3/d. Table 2 summaries the
estimates for condensate and water/MEG content in the
pipeline, given different ways of estimating the condensate
rate. Table 2 also shows simulation results for the condensate
and water/MEG content, where the condensate carry over and
water/MEG injection have been adjusted to match the
measured production rates. The difference in measured and
simulated condensate content for 11 MSm3/d is less than 2%
when comparable assumptions are used for the condensate
production, while for 8 MSm3/d the simulation shows an over-
prediction by 12%.
The aqueous phase content in the pipeline predicted by the
simulation is between 40 and 50% of the measured value for
both flow rates. However, the water/MEG content is only
approximately 5% of the condensate content.
There is a 5 to 10% difference between the condensate rate
estimated from the average export rate and from the tracer
measurements. The results also show that the initial
assumption of 60% condensate carry over at 11 MSm3/d is
somewhat high, 45 to 50% seems more correct.
Table 2. Estimates of condensate and water hold-up from tracer
measurements. Numbers in parenthesis are simulation results for
condensate and water/MEG content.
Rate Condensate Water/MEG Condensate Water/MEG
[MSm3/d] Rate [kg/s] Rate [kg/s] Cont. [m3] Cont. [m3]
11(a) 9.4 0.099 1115 43
11(b) 10.1 0.116 1221 50
(1188) (24)
11(c) 8.6 0.099 993 43
(997) (21)
8 (a) 5.3 0.081 913 82
8 (b) 5.2 0.083 892 84
(996) (36)
8 (c) 5.0 0.081 849 82
(955) (35)
a) Condensate flow estimated from export rate.
b) Condensate flow based on assumption of 60% condensate
carry over at 11 MSm3/d and 40% at 8 MSm3/d. The
water/MEG rate is based on assumed injection rate.
c) Condensate flow determined by dilution method.
The uncertainty in liquid content measurements is always
very large, since the operations involved in collecting the data
are complex, and the end result depends on a series of
measurements and calculations that each has uncertainties.
Typically a 20% uncertainty is assumed in the total
pipeline liquid content, with larger uncertainty in the split
between condensate and water. For the current case the
uncertainty in water content is very large, but the numbers are
also small compared to the total liquid content in the pipeline.
The water content is more interesting at low flow rates where
the water content is much higher.
Shut-in, start-up and surge waves
To illustrate the transient behavior of the pipeline, measured
data and simulation results are presented for start-up of the
pipeline following a shut-in that lasted only a few hours.
Figure 12 shows the gas rate from Huldra, starting from a
steady state rate of almost 11 MSm3/d, a gradually reduction
over 2 hours followed by 3 hours with no production.
Production is brought back up in three steps over
approximately one day, 8 MSm3/d, 9.5 MSm3/d and the finally
11 MSm3/d.
Figure 13 shows the measured pressure at Heimdal, which
increases from 100 bara at the initial steady state, up to a peak
of approximately 125 bara during the shut-in and then
gradually back to 100 bara when production is restarted. Note
that there is a lack of data for a short period at the beginning
of the shut-in, and some assumptions had to be made to make
the data during this period consistent. The measured values for
Huldra flow rate and Heimdal pressure are the input
parameters to the simulation model, although not with all the
details, as shown in Figure 12 and Figure 13.
A comparison of the measured and simulated pressure at
Huldra is also given in Figure 13. There is very good
agreement and the maximum deviation is 2 bar.
Figure 14 shows a comparison of the measured and
simulated gas rates into Heimdal. In the simulation the flow
stops earlier at the beginning of the shut-in, which is a period
were the measured data is not reliable (does not correspond to
other measured values such as pressures and condensate rates).
Further, the gas flow starts earlier in the simulation. This has
partly to do with the assumptions made during the period
where pressure data were not available, but in addition all the
details of how the platform was operated are not available
(opening and closing of valves etc), which will influence the
results of the simulations. However, there is overall very good
agreement.
OTC 17894 7

Figure 15 shows the measured and simulated condensate

rate into Heimdal. At steady state some minor fluctuations are 135

seen in the measurement while the model predicts steady flow. 130

In most cases this has no practical consequence. Following the 125

start-up the initial liquid surge is predicted somewhat later

Pressure [barg]
120
then what is measured. As for the gas rate this is most likely Huldra Export Pressure Measured
Heimdal Import Pressure Measured
caused by lack of information on the details of how the 115
Huldra Export Pressure - Simulated
platform is operated. Following start of production the overall 110
Heimdal Outlet Pressure - Simulation input

features of the simulation match the measurements well, but

105
there are deviations in the details, as could be expected.
100

A challenge that has been encountered at Heimdal is liquid 95

22/1 23/1 24/1 25/1
surge waves, which were not predicted accurately. These Date
waves are typically experienced during the first days of Figure 13. Measured pressure at Huldra and Heimdal, simulation
production after a start-up or after a rate increase. The liquid input for Heimdal pressure (approximated to the measured
flow rate will vary significantly with typically a one hour pressure) and simulated Huldra pressure.
period, as seen for the first day after the start-up in Figure 15.
From the figure we can see that the current model predicts 12

11
these waves quite well in the first period after the start-up, but
10
in reality they continue longer than the model predicts. The 9
main reason for this being a challenge at Heimdal is the very 8

small receiving vessel (7 m3) and the limited process capacity. 7

Heimdal Flow Rate Measured

[MSm3/d]
Heimdal Flow Rate Simulated
In most on-shore systems, with a large receiving vessel, these 6

5
surge waves would most likely not even have been noticed.
4

3
Since the current model does not predict the details of 2

these surge waves well, these are studied as part of the 1

Horizon research project, of which Statoil is one of the 0

22/1 23/1 24/1 25/1
sponsors. At present is seems likely that, during start-up, the Date

waves are caused by liquid that has accumulated in low spots Figure 14. Measured and simulated gas rate at Heimdal.
in the pipeline during shut-in, which during start-up travels
through the pipeline as surges much further than the model 60
Measured Simulated
predicts. The numeric of the current model tends to smoothen
Heimdal Condensate Flow Rate [Tonnes/h]

the surges out as they travel through the pipeline, which 50

explains why the initial surges are well predicted, while the 40
later surges, that have travelled much farther through the
pipeline, have longer period and lower amplitude in the model 30

than in real life. During rate changes the cause of the surge is
most likely large changes in liquid hold-up in certain uphill 20

sections of the pipeline, which can occur even for a relatively

10
small change in the flow rate. Based on current research future
models should be able to better predict when these surge 0

waves will occur and be able to track them through the 22/1 23/1 24/1 25/1
Date
pipeline. Figure 15. Measured and simulated condensate rate at Heimdal.

Blow-down and start-up

12
11
10
In connection with a pigging operation of the pipeline there
9 was concern that there were hydrates in the pipeline, based on
8
Huldra Flow Rate Measured
observations of increasing pipeline pressure drop. In
7
connection with a shut-down of the Heimdal platform for
[MSm3/d]

Huldra Flow Rate Simulation Input

6
5
routine maintenance it was decided the pipeline should be
4
blown down so that any hydrates could melt before the
3 pigging operation was performed. Since there was significant
2 condensate and water in the pipeline and the liquid handling
1
capacity during blow down is limited, simulations were
0
22/1 23/1 24/1 25/1
performed to determine how the blow-down and start-up
Date should be performed.
Figure 12. Measured gas rate from Huldra and simulation input
gas rate from Huldra.
8 OTC 17894

During a blow-down production is initially stopped at
Huldra, while production continues into Heimdal until the
separator pressure reaches 93 bara. The flow is then routed to
the flare system, which is designed for a rate of 12 MSm3/d,
but the blow-down rate from the Huldra to Heimdal pipeline is
limited to approximately 5 MSm3/d (due to potential
simultaneous blow-down of multiple pipelines).
A blow-down simulation has been performed from an
initial steady state based on a flow rate of 6.2 MSm3/d with
20% condensate carry over in the Huldra scrubber. This was
anticipated to be the flow rate when the blow-down was to be
performed.
Although there are no accurate measurements from the
blow-down, the overall impression is that the simulations
agreed well with the observations. The only period where
consistent data are available is for the first 6 hours. Figure 21
and Figure 22 show the measured and simulated pressure and
gas rates at Huldra and Heimdal, where the rate at Heimdal
and pressure at Huldra are taken as inputs to the model. There
is very good match on the Huldra pressure prediction and
reasonably good prediction of the gas rate into Heimdal.
Most importantly there was no significant liquid
production during the blow-down, confirming the main
finding of the simulations.

Figure 16 shows the pressure at the pipeline inlet at 120

Huldra
Huldra, the pipeline outlet at Heimdal and in the flare knock- Heimdal - Flare knock-out drum
Heimdal - Pipeline outlet
out drum during the blow-down. Production at Heimdal 100

continues until the separator pressure is 93 bara, which is

approximately 1 hour 50 minutes after the shut-in at Huldra. 80

Flow is then routed to flare, and it takes approximately 60

hours for the pressure to reach 7 bara (considered a safe

[bara]
60

pressure for melting of hydrates). Notice that the pressure at

Huldra closely follows the pressure at Heimdal, showing that 40
there are no significant liquid blockages of the pipeline.
20
Figure 17 shows the gas flow rate to flare. The peak rate at
the beginning of the blow-down is 4.5 MSm3/d, and the blow- 0
down performed here is the fastest blow-down possible. 0 10 20 30 40 50 60
Time [Hours]

Figure 18 shows the accumulated total liquid volume flow Figure 16. Pressure at pipeline inlet, outlet and in flare knock-out
drum during the blow-down.
out of the pipeline and into the flare knock-out drum. The total
5
liquid volume produced during the blow-down is in the order
of 44 m3, which should be well within the liquid handling
capacity of the system. Approximately 5 hours after the start 4

of the blow-down there is a 10 to 15 m3 slug exiting the

pipeline, which can be handled by the flare knock out drum.
3
The difference in liquid production out of the pipeline and into
[MSm3/d]

the flare knock-out drum is due to flashing caused by the drop

in pressure and temperature over the blow-down choke. For 2

the current case this is actually most of the liquid produced

into the flare system 1

Figure 19 shows the accumulated water volume flow to

flare. Overall the total water production is approximately 0.5 0
0 10 20 30 40 50 60
m3, which causes no liquid handling problem. Time [Hours]

Figure 17. Gas rate to flare during the blow-down.

Figure 20 shows the total liquid content and the water
content in the pipeline. The water content remains almost
constant while the condensate content in the pipeline is
reduced by almost 400 m3. Considering that almost no liquid
enters the flare system this may be a bit surprising. The main
reason is the change in composition of the condensate phase
when the pressure is reduced, causing significant amounts of
condensate to flash to the gas phase. Although this has little
impact on the liquid handling during blow-down it does cause
the blow-down to be slower than anticipated since there is
more gas. The opposite effect is seen when restarting
production since the liquid content has a similar increase when
the pressure is increased, causing a slower build-up than
anticipated.
OTC 17894 9

50 115
Huldra Export Pressure - Measured
Heimdal Import Pressure - Measured
45
Huldra Export Pressure - SImulation
Heimdal Import Pressure - Simulation input
40 110
Total liquid - Pipeline outlet
Total liquid - Flare knock-out drum
35

30 105

[bara]
[m3]

25

20 100

15

10 95

0 90
0 10 20 30 40 50 60 20:00 21:00 22:00 23:00 0:00 1:00 2:00
Time [Hours] Time

Figure 18. Accumulated total liquid rate at pipeline outlet and into Figure 21. Measured pipeline inlet and outlet pressure and
flare system during the blow-down. simulated pipeline inlet pressure. In the simulation model the
0.5
pipeline outlet pressure was approximated to the measured
pressure.
7
Heimdal Flow Rate - Measured
0.4 Huldra Flow Rate - Measured
Huldra Flow Rate - Simulation Input
6 Heimdal Flow Rate - Simulation Result

0.3 5
Water - Pipeline outlet
[m3]

Water - Flare knock-out drum

4
[MSm3/d]

0.2

0.1
2

0.0 1
0 10 20 30 40 50 60
Time [Hours] 0
20:00 21:00 22:00 23:00 0:00 1:00 2:00
Figure 19. Accumulated total liquid rate at pipeline outlet and into Time
flare system during the blow-down.
1200
Figure 22. Measured pipeline inlet and outlet gas flow rate and
Total liquid Water simulated pipeline outlet gas rate. In the simulation model the
pipeline inlet flow rate was approximated to the measured flow
1000 rate.

800 Tail end production

[m3]

600 In connection with tail end production from Huldra the flow
rate has been decreasing rapidly from the beginning of 2004,
400 as shown in Figure 5, and by mid 2005 it had reached a level
were there was no longer continuous water/MEG flow into
200 Heimdal due to increasing liquid accumulation in the pipeline
with reduced flow rate. This has led to hydrates at Heimdal
0 upstream of the inlet separator and problems for the MEG
0 10 20 30 40 50 60
Time [Hours]
regeneration plant. For the Huldra platform this means that
one no longer can base the operation on MEG regenerated at
Figure 20. Pipeline total liquid and water content during the blow-
down. Heimdal. As discussed later a compressor will be installed at
Huldra in late 2006, but it is desirable to maintain production
with the current facility until the compressor is installed.

A suggested solution is cyclic packing/unpacking the

pipeline. This means producing at a constant rate from Huldra
while the production to Heimdal is intermittently low (or zero)
and high (possibly up to 10 MSm3/d).

It is shown that by cyclic packing/unpacking the pipeline

the total liquid and water/MEG accumulation in the pipeline
10 OTC 17894

can be controlled. Water/MEG is pulled out of the pipeline
due to the high flow rate in the unpacking phase, and the
overall water/MEG and total liquid content in the pipeline is
closer to the steady state value at the unpacking rate into
Heimdal than to the value for the lower steady state rate from
Huldra. When unpacking the pipeline the flow rate close to
Heimdal will be high, but the flow rates close to Huldra will
be closer to the average production rate. This means that the
effect of the dynamic pigging is limited close to Huldra.
MSm3/d for the period we are considering. This means that a 3
hour production cycle would most likely result in significant
loss of production due to the increase in average pipeline
pressure, while a 24 hour production period would most likely
have very little impact on the production.
120
Huldra
118 Heimdal

116
Simulations have been performed where the production
rate at Huldra has been kept constant at 4 MSm3/d, while the 114

production rate into Heimdal has been changed cyclically. The 112

[bara]
pipeline has been packed until the pressure at Heimdal reaches 110
a certain level (115 bara). Production into Heimdal is then 108
started and maintained at a constant rate until the pressure
106
drops to 100 bara. Production into Heimdal is then maintained
104
to keep the pressure at 100 bara until the end of a 24 hour
cycle. 102

100
Figure 23 shows the pressure at Heimdal and Huldra. The 0 1 2 3 4
Time [days]
pressure at Heimdal varies from 100 to 115 bara, while the
pressure at Huldra varies between 106 and 118 bara. The Figure 23. Pipeline inlet and outlet pressure during cyclic
production with 24 hour period.
average pressure at Huldra is 108.5 bara, which is only
slightly higher than the steady state pressure at Huldra for 4
MSm3/d production. Figure 24 shows the gas rate from Huldra 10
Huldra
Heimdal
and the rate into Heimdal. Notice that the flow into Heimdal 9

drops gradually from 10 MSm3/d to 4 MSm3/d following the 8

initial unpacking at 10 MSm3/d. This is due to a gradual 7

unpacking of the pipeline.
6
[MSm3/d]

5
Figure 25 shows the pipeline total liquid content and water
content. The variation in water content is very small and the 4

variation in total liquid content is in the order of 60 m3. This 3

variation is partly due to the pressure variations in the 2
pipeline.
1

0
The accumulated liquid flow into Heimdal is shown in 1 2 3 4
Figure 26. Notice that the liquid rate when production is Time [days]
started into Heimdal is high, causing potential liquid handling Figure 24. Gas flow from Huldra and into Heimdal during cyclic
problems. Slugs in the order of 10 to 15 m3 arriving over a few production with 24 hour period.
minutes must be expected at Heimdal, during the start of each 700
production cycle.
600

In addition to the simulation with 24 hour period a

500
simulation was performed for 3 hour cycle (new cycle is Total liquid
Water
started immediately when the Heimdal pressure reaches 100 400
bara) and 48 hour cycle. Table 3 summarizes the main results
[m3]

of the simulations. In addition the results for a steady state run 300
with 4 MSm3/d from Huldra is included.
200

The simulation results show that cyclic production can

100
drastically reduce the condensate and water content, compared
to steady state with the same average rate, without 0
significantly increasing the pipeline inlet pressure. The general 1 2 3 4
Time [days]
trend is that shorter production cycles give lower liquid
content in the pipeline, but also higher average pipeline inlet Figure 25. Pipeline total liquid and water content during cyclic
pressure. A reservoir simulation was performed where the production with 24 hour period.
pipeline inlet pressure was increased by 5 bar, showing a
decline in the average flow rate from the well by 0.25 to 0.5
OTC 17894 11

500 1600
Total liquid Total liquid
450 Water Water
1400

400
1200
350

1000
300
[m3]

[m3]
250 800

200
600

150
400
100

200
50

0 0
0 1 2 3 4 0 30 60 90 120 150 180 210 240 270 300 330
Time [days] Time [days]

Figure 26. Accumulated total liquid and water volume flow at Figure 27. Pipeline total liquid and water content for case with
3
pipeline outlet during cyclic production with 24 hour period. reduction in production from 4 to 2 MSm /d over 330 days.
3 450
Table 3. Summary of results for 4 MSm /d production from Huldra, Water
3
10 MSm /d maximum production into Heimdal and 115 bara 400
packing pressure.
350
Average Average Average
Huldra liquid water/MEG 300

pressure content content 250

[bara] [m3] [m3] [m3]
200

Steady state 107.5 1013 499 150

Cyclic 3 hours 115.4 396 128
100
Cyclic 24 hours 108.4 621 164
Cyclic 48 hours 107.6 712 199 50

0
0 30 60 90 120 150 180 210 240 270 300 330
A simulation has been performed where the production Time [days]
rate at Huldra has been reduced from 4 MSm3/d to 2 MSm3/d
over a 330 day period (approximately as it will decline Figure 28. Accumulated water/MEG flow into Heimdal for case
3
with reduction in production from 4 to 2 MSm /d over 330 days.
according to the production profile), while the production rate
into Heimdal has been changed cyclically. The pipeline has
Extending the field life further
been packed until the pressure at Heimdal reaches 115 bara,
production into Heimdal is then started and maintained at a
As previously discussed the production rate from Huldra in the
constant rate (10 MSm3/d) until the pressure drops to 100 bara,
current configuration is rapidly declining and can not be
when production into Heimdal is stopped and the cycle
maintained after 2006. Since the Heimdal process receives
repeated.
production from a number of fields reducing the pressure at
Heimdal is not a realistic alternative. Different alternatives
Figure 27 shows the pipeline total liquid content and water
were considered for recovering the remaining reserves in the
content. The water content increases from approximately 100
Huldra field [2]. Compression at the Huldra platform was
m3 to 400 m3, while the total liquid content increases from 400
found to be the best solution, and a compressor will be
m3 to 1200 m3 over the 330 day period. The total liquid
installed by the end of 2006.
content also varies approximately 60 m3 over each cycle.
By installing a compressor at Huldra the wellhead pressure
Figure 28 shows the accumulated water/MEG production
can be brought down and significant parts of the remaining
into Heimdal during the entire period. Water/MEG is
reserves can be recovered. From a flow assurance point of
produced into Heimdal during the entire period, even though
view one interesting aspect of the operation of the pipeline
the water/MEG content in the pipeline increases significantly
after installation of the compressor is that flow now will be
towards the end of the 330 day period. This means that there
two phase, gas and water/MEG. Since more condensate is
will be continuous inhibition of the entire flowline, and MEG
extracted from the fluid at Huldra at low pressure, no
will be available for regeneration, although in somewhat
condensate will condense in the pipeline. Further, cooling the
limited quantities.
fluid as much as possible at Huldra minimizes the water
condensation in the pipeline. However, to avoid hydrates
MEG will still have to be injected, which does introduce some
additional liquid into the pipeline.
12
Simulations have shown that the steady state water/MEG
content in the pipeline will be significant. For instance, for a
flow rate of 2 MSm3/d the total water/MEG content is
approximately 3000 m3. However, since the accumulation of
liquid is very slow, due to little water condensation in the
pipeline and low injection rates, the steady state value will
never be reached for production rates beyond some point in
the production profile. For instance, after 15 months of
production it is anticipated that the liquid content in the
pipeline will only be 50% of the steady state value for the flow
rate at the time, and there will be very little liquid in the last
40 km of the pipeline.
In order to limit the liquid in the pipeline it is planned to
pig the pipeline when the liquid content reaches 750 m3. It is
anticipated that it will only be required to pig the pipeline two
to three times during the period when compression will be
used.
On-line monitoring
Statoil has on a number of fields installed on-line pipeline
monitoring systems. The first installation was at Troll [4]
followed by Huldra [3]. Snøhvit will also have such a system
when production starts in 2007.
These systems are based on a pipeline simulation model
being feed real time data for such variables as inlet flow and
temperature and outlet pressure. The model can in addition be
slowly tuned to match the measurement data by changing
selected parameters, such as pipe roughness to match pipeline
pressure drop and ambient temperature to match pipeline
outlet temperature. Based on this the simulation model can in
real time predict pipeline hold-up, pressure and temperature
profiles. This information can be used to predict hydrate
formation and blockages, leak detection and slugging. Further,
a built-in look-ahead model can perform simulations starting
from the current pipeline condition testing out operations that
are to be performed. This is used to verify liquid surges during
rate changes and liquid slugs produced by pigging operation,
significantly reducing the risk of process upsets when the
operations are being performed.
An additional benefit of these on-line systems is that the
close coupling between the simulation model and the pipeline
operation can feed data back into improvements of the flow
models. Either directly from finding cases where the
simulations deviate more from the observations than what is
anticipated or by the operational data which can be collected
in formats that can easily be used for later comparison with
improved versions of the simulation models.
Conclusions
Statoil is operating a number of gas condensate pipelines,
where the application areas range from untreated wellfluids
from subsea tieback to rich gas with limited condensation in
the pipeline.
OTC 17894
The Huldra field was developed with a 150 km long 22”
pipeline to the Heimdal platform. The inlet facility at Heimdal
has a very small receiving vessel (7 m3), and limited liquid
processing capacity. This meant that the design relied heavily
on the accuracy of the design tools used.
The pressure drop in the pipeline has been found to be in
line with the predictions made, but during initial production
the liquid accumulation was higher than predicted by
modeling. Investigations were carried out in order to explain
the observations, and it was determined that condensate carry
over in the Huldra scrubber had a significant influence on the
condensate content in pipeline. Tracer measurements were
performed to determine the condensate content in the pipeline,
which were found to be within 10% of the model predictions
when the correct assumptions for liquid carry over were used.
For the water/MEG phase the predicted content is only 40 to
50% of the measured value. However, the water/MEG content
is very low, only 5% of the condensate content and only 0.2%
of the pipeline volume, were the models used are not expected
to be very accurate.
It has been shown that transient events, such as shut-in,
start-up and blow-down, can be modeled with excellent
agreement between the measured data and the modeling
results. However, a problem that has been encountered at
Heimdal is that, while liquid surge waves are well predicted in
the first period after the start-up, they are experienced for a
longer period than model predicts. Most likely such waves can
in reality travel farther through the pipeline than model
predicted. The reasons for this being a challenge are the
limited liquids buffer and handling capacities at Heimdal. In
most other systems, with a larger receiving vessel, these surge
waves would most likely not even have been noticed.
The field has entered tail end production and is now
operated below the initial design range. Simulations are used
to determine how the field life can be extended even further.
Acknowledgements
The authors acknowledge the Huldra license with the partners
Petoro, Total, ConocoPhillips, Statoil and Paladin Resources
for the permission to publish this paper. Norsk Hydro as the
operator of Heimdal is also acknowledged.
References
1. Inyang, S. E., Tak, A. N. H., Costello, G.: Optimization of a large
integrated area development of gas fields offshore Sarawak,
SPE 29306-MS, SPE Asia Pacific Oil and Gas Conference, 20-
22 March 1995, Kuala Lumpur, Malaysia.
2.Hagesæther, L. et al.: “Rich gas pipeline operations during tail-end
production”, paper PSIG 0402, presented at the 2003 PSIG
Annual Meeting, Palm Springs, California, USA, October 20-
22.
OTC 17894 13

3.Postvoll, W. et al.: “Huldra: Initial Experience in Real-Time

Multiphase Pipeline Modelling”, paper PSIG 0209, presented at
the 2002 PSIG annual meeting, Portland, Oregon, USA, October
23-25.

4.Klemp, S., Meland, B., Hustvedt, E. and Østby, O.: “Operational

experiences from multiphase transfer at Troll”, presented at the
8th International Conference on Multiphase ’97, Cannes, France,
June 18-20 1997.

5. Gilchrist, R.T. and Kluwen, F.A.: Mensa Project: Flowlines, OTC

Paper 008628, Presented at Offshore Technology Conference
1998, May 4-7 1998, Houston, Texas.

6. Wallace, B.K, Gudimetla, R., and Saether, G.: Canyon Express

Slugging and Liquids Handling. OTC Paper 013073, Presented
at Offshore technology Conference 2001, Houston, Texas,
April 30 – May 3 2001.

7. Cooley, C., Wallace, B.K., and Gudimetla, R.: Liquid Management

on Canyon Express, SPE 84346-MS, Presented at SPE ATCE,
Denver, Colorado, 5-8 October 2003.

8. Harun, A.F., Cochran, S.W. and Choate, T.G.A.: Optimization of

Ramp-Up Operations of a Dual Large Diameter Long-Distance
Subsea Wet Gas Pipeline System, SPE 77652, Presented at SPE
ATCE, San Antonio, Texas, 29 September - 2. October 2002.

9. Lingelem, M.N., Braseth, A.O., and Holm, H,: Experience from

Multiphase flow operations and pipeline operation on TOGI,
Presented to the Norwegian Petroleum Society Conference on
Multiphase Transport, Røros, Norway, 20-22 September 1992.

10. Wilson, A., Overaa, S.J, and Holm, H.: Ormen Lange - Flow
assurance challenges, OTC Paper 016555, Presnted at Offshore
Technology Conference 2004, Houston, Texas, May 3-6 2004.

11.Hærdig, A., Moe, R., Klemp, S. and Ravnøy, J.: ”Transients in

network pipelines – problems related to tie-ins between already
existing flowlines and new ones”, presented at the 9th
International Conference on Multiphase ’99, Cannes, France,
June 16-18 1999.

12.“Fact Sheet 2005 Norwegian Petroleum Activity”, Norwegian

Ministry of Petroleum and Energy, Postboks 8148 Dep, 0033,
Oslo, Norway (2004).

13. Eidsmoen, H. and Roberts, I.: Issues relating to proper modeling

of the profile of long gas condensate pipeline, PSIG 0501,
Presented at PSIG Annual Meeting, San Antonio, Texas,
November 7-9 2005.

14. Sira, T., et al.: Liquid accumulation in gas condensate pipelines

measured by use of tracer techniques. Presented at 12th Int.
Conference on Multiphase Production Technology ’05, BHR
Group, Barcelona, Spain, May 25-27, 2005.

15. Sira, T. and Flaten, G. (Statoil). Application on patent on “Tracer

measurement of phase volume in multiphase pipelines”. Patent
applications “NO 20045642” and “PCT NO2005/000421”.