Professional Documents
Culture Documents
net/publication/254535636
Horizontal Openhole Gravel Packing in Wells With Long Blank Pipe and
Screen Sections
CITATION READS
1 426
5 authors, including:
6 PUBLICATIONS 11 CITATIONS
University of Stavanger (UiS)
37 PUBLICATIONS 308 CITATIONS
SEE PROFILE
SEE PROFILE
Arild Saasen
Universiy of Stavanger
334 PUBLICATIONS 2,729 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Yohanes Ronny Costamte Siregar on 30 October 2020.
Horizontal Openhole Gravel Packing in Wells with Long Blank Pipe and
Screen Sections
1
Y.R. Costamte, SPE, University of Stavanger and Statoil ASA; A. Trondsen, SPE, R. Bergkvam , Halliburton;
H. Hodne, University of Stavanger; and A. Saasen, SPE, University of Stavanger and Det norske oljeselskap ASA
This paper was prepared for presentation at the SPE Deepwater Drilling and Completions Conference held in Galveston, Texas, USA, 20–21 June 2012.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers 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 SPE copyright.
Abstract
For offshore developments it is anticipated that several wells will be completed using gravel packed sand screens divided by
long blank pipe sections. Open-hole gravel packing has routinely been used as a sand control method in long horizontal wells.
However handling long blank pipe sections represents a challenge. The length of blank pipe section could be 600 meters or
longer. In such cases, the chart of the surface pump pressure versus the pumping time is different when compared with a
typical profile from a horizontal gravel pack in a section without blank pipe. Typically, gravel pack sand moves in from heel to
toe of a well for roughly half of the cross section area (alpha wave) followed by a final placement of the remaining part of the
annulus from toe to heel (beta wave).
Since there is no flow through the pipe in the blank pipe section, the beta wave does not exist here, and its absence
complicates the gravel placement in the heel region of the reservoir. On the upstream side of the blank pipe, the slurry is
dehydrated through the screen. This slows down the rate in the area adjacent to the blank pipe where in turn a new alpha wave
adjacent to the blank pipe starts. This process appears to continue until the majority of the area open to flow above the alpha
wave adjacent to the blank pipe has been filled in. The beta wave adjacent to the screen on the upstream side of the blank pipe
then continues normally.
The paper describes in detail a mechanistic model based on experiments which predicts the phenomena discussed above.
This model for gravel placement has been validated both qualitatively and quantitatively with both laboratory and the North
Sea field Heidrun data.
Introduction
The completion intervals for horizontal wells are becoming longer and complex. Horizontal drilling, combined with reliable
open-hole completions, has an advantage in that more reservoir rock is exposed to the wellbore when compared with vertical
or deviated wells and cased-hole completions. If a sand production problem occurs, Open-Hole Gravel Pack (OHGP) has
routinely been used as a sand control method in horizontal wells. A lot of efforts in the industry have been made to separately
understand parts of the gravel pack placement process. Several authors present empirical results from experimental
observation performed in small scale and full scale field models like Gruesbeck et al. (1979), Forrest (1990) and Harestad et al.
(1991), Penberthy and Echols (1993).
Later, Nguyen et al. (1992), Sanders et al. (2002), Ojo et al. (2006), and Jain, S. et al. (2008) are some of the authors who
built simulators based on empirical models and solve the fundamental laws of physics describing the conventional alpha and
beta wave process.
Most of the available models are only considering OHGP with a screen along the entire horizontal well. Some operators
often install a long blank pipe section between two screen sections, since they are only targeting selected zones with potential
for sand production. Statoil has used this method since the beginning of the last decade, recognising that gravel also would fill
the blank pipe section and provide zone isolation between the two screen section zones (Fjellstad 2012). The length of the
blank pipe section could be long, sometime even longer than 600 m. A field record of the surface treating pressure during the
pumping time period shows an anomalous profile when compared with the existing non-blank pipe pressure and pump rate
1
Now with A/S Norske Shell
2 SPE 150546
profiles from the alpha-beta model. An anomalous profile is suspected to be due to the long blank pipe section where the
normal beta wave treating pressure profile does not occur after the alpha wave placement.
a 1
2
Equilibrium
b 1 dune
height
3
2
3 2
4a
3 2
Option 1
3 2
1
Packing
4b mode
3 2
Option 2
4 3 2
5
4 2
3
Heel Toe
Flow direction of slurry or carrier fluid
Fig. 1–Gravel placement sequences in a case with long blank pipe section in the middle of horizontal section.
In a process of building a gravel placement model, the alpha (wave) dune height becomes a critical parameter. Chen (2007)
compared three different particle transfer and critical velocity models that are usually used by the industry, their applications in
alpha (wave) dune height calculations, their limitations and effective application windows. In the following sections, an
SPE 150546 3
updated model is presented. This model includes the presence of a long blank pipe section in the middle of the horizontal
section as shown in Fig. 1. The model results are compared first with a small scale series of experiments, then with a North
Sea field case.
Beta wave
New alpha waves (at heel screen)
Surface treating pressure, bar
Slurry stage
Screen out
Fig. 2–Typical surface treating pressure profile of an open hole gravel pack operation for a horizontal well competed only with screen
and in case of long blank pipe section in the middle.
After the alpha process is finished, the bed will move backward like a piston, packing all the space above the alpha wave.
This is known as the beta wave. When the beta wave grows, the slurry flows inside the annulus will drop the gravel and the
carrier fluids will flow through the screen into the mini-annulus between the screen basepipe ID / washpipe OD as determined
by friction pressures. As the gravel is deposited in the area above the alpha wave the bed moves backward to the heel of the
well. Once the beta wave has reached the top of the screen the open hole ID/screen OD annulus is assumed to be filled with
gravel. The surface treating pressure (pump pressure) profile during operational time is shown in Fig. 2. By using this pressure
profile and by using mass balance calculations, the end result of the operation can be evaluated and interpreted. The data
usually shown from the operation are pumping pressure, pumping rate, amount of slurry and slurry density. Those data are
plotted with respect to operational time.
When the slurry starts to flow out of the X-over tool and fills up the horizontal section, the alpha wave is building and the
dune is moving toward the toe of the well. The alpha dune height is a function of actual velocity and critical velocity. The
alpha wave will continue until it reaches the toe as long as the conditions remain the same (constant wellbore ID, constant
fluid losses to the formation and to the screen basepipe ID/washpipe OD). The slurry flows inside the annulus and the carrier
fluid flows inside the mini-annulus (Fig. 1, Step 2).
Once the alpha dune reaches the toe screen, the beta dune starts to build up (Fig. 1, Step 3). Since the gravel cannot flow
through the screen, it is deposited in the area open to flow on top of the alpha dune. The carrier fluid leaks off into the screen
leaving the gravel behind in the area open to flow above the top of the alpha wave.. This process continues but the process
slows down when the beta dune reaches the lower end of the blank pipe section as there is no opening in the blank pipe section
where the carrier fluid can flow through.
On the upstream side of the blank pipe, the slurry is still being dehydrated through the screen. This slows down the rate in
the area adjacent to the blank pipe which in turn causes a new alpha wave to start with the losses occurring into the screen
section at the bottom end of the blank pipe section thus causing additional pressure drop. Once the new alpha wave height
reaches the bottom of the screen more fluid is dehydrated through the screen section above the blank pipe thus reducing the
rate and causing another new alpha wave adjacent to the blank pipe section. (Fig. 1, Step 4a and b). However, the new alpha
wave does not form above the heel screen because the flow above it still has the same rate as at an equilibrium condition.
The concept of how the new alpha wave takes form is inspired by Calderon et al. (2007) in designing multiple-alpha waves
on purpose during OHGP operations. The study suggests decreasing the flow rate once the first alpha wave reaches the toe
section, which creates a repeated new and higher alpha wave on top of the first alpha wave. Throughout this paper we will
attempt to address the question of how many new alpha waves could be made to form above the blank pipe section as the
second alpha wave dune does not fill up the entire flow path or void above the first alpha (wave) dune at the blank pipe
section. A small flow path of annulus will still be open for the slurry to flow, and then a new alpha wave will be able to move
from heel to toe of the blank pipe section. The verification of this process will also be described in this paper (Fig. 1, Step 4a
and b).
Larsen (2010) investigated how the packing mode takes place. The experimental test shows that a dual alpha wave packing
mode occurs after the beta wave reaches the lower part of blank pipe (Step 4b). With regards to simplicity in calculations, the
developed mathematical model follows step 4a.
Once the area adjacent to the blank pipe has been packed as much as possible with dual alpha waves the beta wave above
the top end of the blank pipe continues until the top of the screen at the heel of the well has been reached, following the
concept of step 3. Once the heel screen is covered by the gravel, there is no opening area for carrier fluid to flow. This
condition will end the formation of the beta dune and dramatically increases the pressure drop, and thus creates the so-called
screen out. (Fig. 1, Step 5)
A set of assumptions are needed to build the present model. Detailed explanations for each assumption is described by
Costamte (2010): no rat hole and no wash-out; no pressure drop radially through the screen for the flow from annulus into
mini-annulus at horizontal section; fully eccentric annulus configuration is used for horizontal section and concentric
configuration is used for vertical section; neglected pressure drop due to acceleration; pressure drop due to gravity in
horizontal section is zero; no fluid loss into reservoir; use a Newtonian carrier fluid with a low viscosity; and splitting of flow
rate starts in the cross-over gravel-pack port.
condition is 85oC static and 40-50oC during circulation. Pump rate during operation is 1 100 lpm. Additional field data and
completion schematic is shown in Fig. 3. The well data is used for the alpha (wave) dune height calculation.
Heel Toe
Heel Screen Blank Pipe Toe Screen TD: 4220 mMD / 3458 mTVD
Fig. 3–Completion schematic. 10 gauge screen is used with 22. 490 in2/ft of effective inlet area of wire wrapped screen jackets. OD
screen with wire = 6.05in, OD base pipe= 5.5 in, ID screen= 4.89 in, and OD wash pipe = 4.00 in.
Chen (2007) and Pedroso et al. (2005) discussed the choices for the value of alpha dune height and critical velocity as
described in the next paragraphs.
0.16
Dh
0.14 Deff
Dh (modified)
0.12
0.10
Diameter, m
0.08
0.06
0.04
0.02
0.00
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0
Dune Height, in
Fig. 4–Hydraulic diameter, effective diameter and modified hydraulic diameter with regards to dune height.
3,0
Actual Velocity
Critical Velocity-Gruesbeck
Critical Velocity-Penberthy1
2,5 Critical Velocity-Penberthy2
Critical Velocity-Oroskar & Turian
Avg G and O&T
2,0
Velocity (m/s)
1,5
1,0
0,5
0,0
0 10 20 30 40 50 60 70 80 90 100
Dune (%-height)
Fig. 5–Plotting actual and critical velocity with regard to dune height.
Gruesbeck et al.’s model (1979): The experiment does not include the effects of screen-wash pipe annulus and fluid leak
off; physical model itself was small compared to field scale; and the alpha wave height calculated turned to be slightly higher
than the data collected from actual jobs. Oroskar and Turian’s model (1980): This correlation takes into account both the
hindered settling velocity and turbulent fluid eddy currents within pipe; the correlation was not developed for the purpose of
gravel packing, but for general objectives of particle transfer; and the calculated velocity needs to be modified to match field
data. A model by Shah and Lord which is based on the Oroskar and Turian model has frequently been used to determine
alpha/beta waves and it has been checked in large scale testing and found to reasonably match test results.
Penberthy et al.’s model (1996): The model was conducted in a field scale model and fluid leak off was simulated; the
predicted Alpha and Beta wave were closely matched to the test results; critical velocity may display different trends when
switch from V1 to V2; fluid viscosity and gravel concentration are not in the V2 calculation; and the critical velocity can be
estimated within range of V1 and V2.
By considering the discussion above, the Penberthy et al.’s model can be used as a prediction range for critical velocity
value. Also it can be concluded that the critical velocity value must be less than Gruesbeck et al.’s model. Since only
Gruesbeck et al.’s model and Oroskar and Turian’s model take into account the settling velocity parameter which is a main
aspect in sediment transport, so the value of critical velocity is agreed as an average of those two models. As shown in
Table 1, the value of critical velocity is 1.28 m/s, which give a dune height of 62.7%-height of the open-hole diameter.
SPE 150546 7
4 f ' ρ an v 2 4 f ' ρ ma v 2
=
D g 2
h c an Dh g c 2 ma
Geometry
Vs Vc , H
Dh, A
ΔP
Fig. 6– Flow diagram for calculating flow rate splitting above and in front of dune (Gruesbeck’s).
areas above dune and in front of the dune. This condition gives a different splitting flow rate which is described in detail in
Appendix. A flow rate splitting in front of the dune is using Gruesbeck’s pressure drop gradient balance as shown in Fig. 6.
From the result of the flow rate splitting in between two different flow areas in Table 2, it can be concluded that the flow
rate in annulus above the dune is less than that in front of the dune. It shows that some amount of carrier fluid in the mini-
annulus flows outside to the annulus in the area of front of dune. The flow rate in the mini-annulus is higher than the initial
condition when there is no dune and only carrier fluid is flowing in the annuli. And the flow area above dune in annulus is less
than the initial condition which creates more friction pressure drop. These two reasons cause an increase in pressure drop
gradient when the alpha dune is moving which is shown in surface treating profile in Fig. 7.
The calculation from Eq. 1 gives 202.75 min to pump the slurry until the toe of horizontal section. The speed of the alpha
dune is 0.062 m/second as calculated by using Eq. 3.
Lm (3)
vp =
TP
∆p
Flow rate
L
Rate, m3/min Annulus Mini-annulus Pa/m
Initially (no dune) 0,854 0,246 479,05
Above dune 0,670 0,430 1410,83
In front of dune 0,852 0,248 485,15
Table 3–Opening area and flow rate splitting inside beta dune and toe screen.
OPENING AREA
Screen (10 Gauge) Beta (wave) Dune
2
Screen Opening Area : 0.015 m /ft Gravel Conductivity : 13040 md-ft
% of the real opening area : 24.46 %
2 2
Real Screen Opening Area : 0.004 m /ft 0.0004% Gravel Opening Area : 0.000000013040 m /ft 99.9996%
FLOW RATE
Before Reach Beta Dune (No Beta Wave) Reach Beta Dune (Beta Wave)
3
Annulus : 0.6699178 m /min Splitting of Flow Mini-annulus : 0.669915359 m3/min 99.999633%
Mini-annulus :
3
0.4300822 m /min rate in Annulus Beta Dune : 0.0000025 m3/min 0.000367%
3
Pump Rate : 1.1 m /min Mini-annulus : 1.099997539 m3/min 99.999776%
Final Flow rate
Beta Dune : 0.0000025 m3/min 0.000224%
drop in the vertical section and inside the wash pipe is constant as calculated in step 2 above, but pressure drop in two annuli in
the horizontal section is changed during this pumping time period.
Flow rate in the annulus above the alpha dune before it reaches the end of heel screen is the same as calculated in step 2.
Gruesbeck’s pressure drop gradient balance is used to calculate this flow rate, since it is still in equilibrium condition. Once
the beta wave reaches the lower end of the blank pipe the pressure increases due to the lack of losses through the blank pipe.
This causes an increase in pressure at the lower end of the screen (at the top of the blank pipe) which causes additional losses
to the mini annulus (basepipe ID/washpipe OD) with a subsequent decrease in the flow rate in the annulus (wellbore/screen
and wellbore/blank pipe). The decreased flowrate starts a new alpha wave which propagates to the bottom end of the blank
pipe. Same concept of flow rate splitting in step 3 above is applied in this step 4. The flow rate in the annulus is splitting to an
opening area of heel screen and to annulus above blank pipe. This idea is used as an initial condition to figure out what is the
ratio of flow rate splitting.
Table 4–Opening area and flow rate splitting between annulus above blank pipe and heel screen opening.
INITIAL ITERATION
OPENING AREA
Heel Screen (10 Gauge)
Screen Opening Area (Fig. 5.3) : 0,015 m2/ft
% of the real opening area (Fig. 7.5) : 24,46 %
Real Screen Opening Area : 0,367 m2/ft
Heel screen lenght : 71,70 m
Real Screen Opening Area : 0,835 m2 98,66 % 96,10 %
Flow Area Above Alpha Dune
Flow Area : 0,0113 m2 1,33 % 3,90 %
FLOW RATE
Equilibrium Condition
Annulus : 0,6699178 m3/min
Mini-annulus : 0,430082 m3/min
Pump Rate : 1,1 m3/min
When Splitting
Splitting of Flow Mini-annulus : 0,660973 m3/min 98,66 % 0,643791 m3/min 96,10 %
rate in Annulus Annulus : 0,008910 m3/min 1,33 % 0,026127 m3/min 3,90 %
3
Mini-annulus : 1,091055 m /min 99,19 % 1,073873 m3/min 97,62 %
Final Flow rate
Annulus : 0,008910 m3/min 0,81 % 0,026127 m3/min 2,38 %
175
150
Surface Treating Pressure, bar
125
100
5
4
75
3
50 1 2
25
0
0 60 120 180 240 300 360 420 480 540 600
Time, min
The calculation gives that 1.33 % of flow rate from equilibrium conditions (0.6699178 m3/min) is flowing to the annulus
above the blank pipe which is 0.008934 m3/min. This ratio of flow rate splitting is used as an initial condition for calculation
of surface treating pressure. Result of the calculation of flow rate splitting is shown in Table 4.
The new alpha dune height is calculated by using an initial condition value of flow rate in the annulus above the blank
pipe. An error occurred on the critical velocity of the Oroskar and Turian’s model calculation. This error is occurred because
of an error in gravel concentration term in Eq. A-19, if it is calculated by using the initial condition flow rate splitting value.
Because of this error, next action is to do iteration by changing the ratio of flow rate that flow in the annulus above the blank
pipe. A value of 3.9% give a good result, since the profile of surface treating pressure from the calculation on step 4 can
connect smoothly with the surface treating pressure profile on step 3 and step 5 as shown in Fig. 7 and the calculation of
critical velocity of Oroskar and Turian’s model is not error. A new alpha (wave) dune height is 92%-height (~7.8 in) and 0.304
m/s of critical velocity which shows a similar manner with respect to Fig 6.
10 SPE 150546
Since there is no opening area on the blank pipe section in this step, the flow rate in front of the new-alpha (wave) dune is
the same as above it. But the same flow rate as in step 4 is not happened for the initial alpha-wave in step 2.
Pumping time for step 4 is 87 minutes as calculated by using Eq. 1. The speed of the new-alpha (wave) dune is 2.586m/min
(0.043m/s) based on calculations using Eq. 3.
Fig. 8–Mini model with detail rig up showing each part of the components.
Laboratory experiments
One type of gravel pack test model, which is used as a means to study and observe the gravel placement in the horizontal
section, is a mini model or scaled-down model as shown in Fig. 8. The model is scaled down approximately 2.5 times from
actual well configurations but still retaining the same annulus ratio (A/a) between outer and inner/mini annulus. Fig. 8 shows
that small length of blank pipe installed between the two screens. Detailed model rig data is described by Costamte (2010).
This model is too short in length to explain the whole phenomena for model of horizontal wells with a long blank pipe
section in between two screens section. Nevertheless the height of alpha dune during equilibrium conditions and total packing
efficiency can still be obtained from running these tests. Thus, the height of alpha dune can be used to validate the calculation
SPE 150546 11
result, critical velocity and alpha (dune) wave height, from three different particle transfer and critical velocity models.
Two experiments were run using water as a carrier fluid at temperature 10oC and using 20/40 US Mesh gravel. The screen
in the mini model is laid down on the lower part of the glass outer pipe (fully eccentric configuration) which is represented in
model structure and field condition. Water is used as carrier fluid (Newtonian fluid).
The alpha dune height in the laboratory experiment is measured in %-volume. The packing mode happened as illustrated
in Fig. 1 on step 4b. The beta wave propagates to the top of the toe screen and stops at the bottom of the blank pipe section.
After the packing is finished, the beta wave starts to move again until filling up the whole annulus. The new alpha wave were
not easily observed in this test, the reason being that the mini-model limited blank pipe length will not representing the ideal
field condition for the long horizontal blank sections. A total of 13 data points of alpha dune height was obtained from the
experiment (see Table 5).
100
90
80
Model structure (%-volume)
70
60
50
0
0 10 20 30 40 50 60 70 80 90 100
Experimental (% -volume)
Fig. 9–Correlation of calculated alpha dune height using Oroskar and Turian model, Gruesbeck model and average value of them
versus laboratory experiment results.
100
90
80
Model structure (%-volume)
70
60
50
1270 lpm,108 kg/min
40 1100 lpm,108 kg/min
900 lpm,108 kg/min
30
750 lpm,108 kg/min
1270 lpm,153 kg/min
20
+10%
Fitting
10
-10%
0
0 10 20 30 40 50 60 70 80 90 100
Experimental (% -volume)
Fig. 10–Distribution of calculated alpha dune height data using Oroskar-Turian’s Model by varying pumping rate and slurry
concentration. Most of the data are within 10% reliability region of the laboratory experiment results.
Fig. 11–Profile of field surface treating pressure record during OHGP operation in one of the well at the North Sea Heidrun Field.
Table 6–Validating pumping time and surface treating pressure from the mathematical model with recorded field data.
Model Interpreted Field Record
Step Pump time Treating Pressure (bar) Pump time Treating Pressure (bar)
(min) Interval Delta (min) Interval Delta
1 29,1 40,28 - 33,79 -6,5 32,0 52 - 43 -9
2 202,7 33,79 - 45,81 12,0 343,0 43 - 50 7
3 133,3 45,81 - 66,41 20,6 105,0 50 - 70 20
4 87,0 66,41 - 78,76 12,4 70,0 70 - 88 18
5 31,8 78,76 - 83,39 4,6 44,0 88 - 100 12
screen out - - - - - 100 - 130 -
Total 483,9 594,0
Difference 110,1 min
110
100
90
Surface Treating Pressure, bar
80
70
60
50
40
30
Model structure
20 Field (Real)
Field (modified)
10
0
0 60 120 180 240 300 360 420 480 540 600
Time, min
Fig. 12–Comparison of surface treating pressure profiles from the mathematic model versus interpreted recorded field data.
Comparison of the mathematical model with North Sea Field Heidrun data
Qualitatively, the model of the surface treating pressure profile in Fig. 7 has similar tendency profile with the recorded
field data (see Fig. 11) and also with illustration of the model in Fig. 1. This is shows that the explanation and calculation of
the model can be used to answer the phenomena of alpha-beta wave on screen section and also a new-alpha wave on a long
blank pipe section. The profile of surface treating pressure of the model can not be plotted together with the field surface
pressure record, since the ASCII data is not available.
It is necessary with a quantitative validation to conclude that the model matches the recorded field data. The differences in
period of pumping time and value of surface treating pressure between the mathematical model and recorded field data are
shown in Table 6. The value of recorded field data is interpreted from Fig. 11. Based on this interpretation, plot of model and
interpreted field record is presented in Fig. 12. Referring to Table 6 and Fig. 12, it is easier to compare quantitatively the
mathematical model with the recorded field data.
SPE 150546 13
The mathematical model give the starting value of surface treating pressure at 40.28 bar on step 1 but it shows 52 bar on
recorded field data. This gives the level of surface treating pressure of the model profile lesser for the whole step compare to
the field record as shown in Fig. 12. The model does not take into account the pressure drop before wellhead. Besides that, the
pressure drop due to acceleration, which might contribute to the pressure drop, is also neglected. This is because of many
changes in diameter inside the tubular; this pathway for the carrier fluid or slurry cannot be stipulated to the level of detail
required.
Regarding to the interval of pumping time, the real operation took 595 minutes to complete the job, but the model shows
483.9 minutes which gives a difference of 110.1 minutes. The post job report shows that the amount of pumped gravel was
115% (22.79 ton-weight gravel) of the calculated plan. Based on this report, the suspected reason for the excess of 15% of
gravel are washouts in the open hole section of the wellbore. From a simple calculation, 15% of excess gravel contributes an
extra 85.5 minutes of pumping time.
The main difference of pumping time between model and the interpreted field record is shown on step 2 in Table 6, when
an alpha dune is formed. The difference in step 2 is 140.25 minutes. Jain, S. et al. (2008) concluded that the dune height is not
constant along the horizontal section. Interestingly, the alpha wave height increase with length. This occurs because the longer
the slurry travels, the more time it is exposed to the formation, which increases the leak off and decreases the flow rate. With
low flow rate a smaller cross section is required to reach the critical velocity, resulting in increased dune height. If it is what
happened, the pumping time and also volume of gravel that was pumped into the well might be higher than predicted by the
model. The pack efficiency for alpha wave (step 2) in Table 7 shows that the model gives 37.53%-volume and 62.7%-volume
on field record.
If it is assumed that 15% excess gravel due to a washout as indicated in step 2. Based on this assumption, modification on
surface treating pressure profile on field record can be done as shown in Fig. 12. Pumping time for step 2 in field record is
reduced 85.5 minutes (if there were no washout). Taking all of that into account, the field record profile gives a good similarity
with the model.
As illustrated in Fig 3 in step 4a, the new-alpha waves results in an increased pressure drop during gravel placement
adjacent to the blank pipe section. This is because of the flow rate mostly flows inside the mini-annulus. But the result in Table
6 on step 4 shows big difference in the delta treating pressure compared with the interpreted field record. The mathematical
model gives 12 bar compared to 18 bar for the interpreted field record. The illustration in Fig. 1 in step 4b can answer this
difference. After the area open to flow above the alpha wave adjacent to the toe screen is filled by the beta wave, the dual
alpha wave method of packing the area open to flow above the alpha wave adjacent to the long blank pipe section begins. The
dual alpha wave packing mode takes a short period of pumping time but results in a huge increase in pressure drop. Since the
mathematical model is following step 4a in Fig. 1 without take into account the packing mode, the pumping time and pressure
drop estimation during step 4a in Fig. 1 still lesser than the field record.
From the beginning, it was mentioned that the packing wave above the existing alpha wave adjacent to the blank pipe
section is a new alpha wave without knowing how many alpha waves are formed. The mathematical model shows that only
one layer which is filling up above it, since the flow rate that flow above the blank pipe reduces only once. The calculation
shows the flow rate at the annulus reduces from the initial flow rate of 0.6699178m3/min (step 2) to 0.026127m3/min (step 4).
Fig. 7 and Fig. 11 also show that the profile during gravel placement at the blank pipe is not repeated. Note: second-alpha
wave might be used as name for a dune above the first alpha wave.
Based on the calculations, the alpha waves above the blank pipe which is included in the new one increase in height (92%-
height of the dune). Some flow area above new-alpha wave is still open. Table 7 shows that the field data gives a pack
efficiency 97.01%-volume for second alpha wave. It means that the blank pipe section is not packing 100%. The open area
above new-alpha wave gives space for slurry to keep flow until the edge of the blank pipe.
The difference in delta treating pressure on step 5 in Table 6 is caused by the packing mode as explained previously. After
the heel screen is filled up by the beta wave, the flow path to the top of the screen (at the heel of the well) is restricted. At this
point the methods of packing switches from a circulation pack to a squeeze pack resulting in a fast increase in pressure drop
(unless the flow rate is reduced). This is called the screen out.
Table 7–Comparison of total packing efficiency from the mathematical model versus recorded field data.
Mathematical Model Field Data
Step
%-height %-volume %-volume
Alpha Wave (2) 62.0 37.53 62.7
New-alpha wave (4) 92.0 92.1 97.01
Conclusion
• A simple and reliable mechanistic model for gravel placement in an open-hole horizontal well completion with
configuration of a long blank pipe section between two screens is presented.
• Penberthy et al. (1996)’s modified hydraulic diameter, which is a function shape factor and effective diameter, can deal
with fully eccentric annuli which eccentricity equal with 1.
• Oroskar and Turian (1980)’s critical velocity model has been found to give a good correlation with the experimental test.
14 SPE 150546
Therefore, Oroskar and Turian’s model was selected to calculate alpha (wave) dune height as a basis for developing
surface treating pressure calculation of the mechanistic model.
• The flow rate split between the annulus (wellbore ID/screen OD) and mini-annulus (basepipe ID/washpipe OD) is
determined from a pressure drop (friction) calculation. The pressure drop gradient balance combined with a new concept of
open area ratio was found to be applicable for calculating the flow rate split.
• Gravel placement over a long blank pipe is different from a placement over the screen. Two-alpha wave dunes (layers)
were found, instead of multiple alpha waves, to fill up the blank pipe section with a packing mode at the toe section of the
blank pipe when before a new-alpha wave is formed.
• Plot of surface treating pressure versus pumping time of the model give a reasonable profile compared with a recorded
field data. This result shows that the model can reveal the dissimilar surface treating pressure from a typical profile of
alpha-beta model.
Acknowledgement
The authors would like to thank University of Stavanger, Halliburton, and Statoil ASA for permission to publish this paper.
The authors also appreciate the contribution from Det norske oljeselskap ASA.
Nomenclature
A Open cross sectional area to flow, L2, m2 [in2]
*
A Cross sectional area of gravel dune, L2, m2 [in2]
ADP Cross sectional area of drill pipe, L2, m2
Ama Open area to flow in mini-annulus, L2, m2 [in2]
C Gravel concentration, m/t, kg/min
CD Drag coefficient
Cv Gravel volume concentration, L3/L3, m3/m3
Ci Initial particle (gravel) volume concentrations, L3/L3, m3/m3
Can Equilibrium particle (gravel) volume concentrations, L3/L3, m3/m3
d Characteristic length
dp Particle (gravel) average diameter, L, μm
dit Outer diameter of inside tube or hole, L, m [in]
dot Inner diameter of outside tube or hole, L, m [in]
Dh Hydraulic diameter, L, m [in]
DS Outer diameter of screen or blank pipe, L, m [in]
DW Diameter of open-hole, L, m [in]
Deff Effective diameter, L, m [in]
Dh-modified Modified hydraulic diameter, L, m [in]
D1 Initial diameter of the open flow area
D2 Final diameter of the open flow area
f’ Fanning friction factor
f Moody friction factor
Fb Buoyancy force
FD Drag force
FG Weight (due to gravitation) force
g Acceleration of gravity, L/t2, m/s2
H Alpha (wave) dune height, L, m [in]
K Permeability
Lm Length of dune movement or Length of assembly section, L, m
NRe Reynolds number
rot Inner radius of outside tube or hole , L, m [in]
rit Outer radius of inside tube or hole, L, m [in]
RS Outer radius of screen or blank pipe, L, m [in]
RW Radius of open-hole L, m [in]
Rep Reynolds particle number
Rep Reynold number defined for Oroskar and Turian’s model
P Wetted perimeter of cross section which contact with the fluids , L, m [in]
Qi Initial volumetric flow rate or pump rate, L3/t, m3/min [lpm]
Qan Volumetric rate over the equilibrium bank (annulus), L3/t, m3/min [lpm]
s Particle (gravel) and liquid (carrier fluid) density ratio
S Shape factor in a relation of effective and hydraulic diameter
TP Pumping time, t, minutes
SPE 150546 15
References
Beggs, H. D. 2003. Production Optimization Using Nodal Analysis, second edition. Tulsa, Oklahoma: OGCI and Petroskills.
Calderon, A., Magalhaes, J.V.M., Oliveira, T.J.L., Martins, A.L. 2007. Designing Multiple Alpha Waves for Openhole Gravel-Pack
Operations. Paper SPE 107615-MS presented at the European Formation Damage Conference, Scheveningen, The Netherlands, 30
May-1 June. doi: 10.2118/107615-MS.
Chen, H. 2008. Flow and Pressure Profile Distributions in Gravel Pack Equipments. MSc. Thesis, Stavanger U., Stavanger, Norway.
Chen, Z. 2007. Horizontal Well Gravel Packing: Dynamic Alpha Wave Dune Height Calculations and Its Impact on Gravel Placement Job
Execution. Paper SPE 110665-MS presented at the 2007 SPE Annual Technical Conference and Exhibition, Anaheim, California, USA,
11-14 November. doi: 10.2118/110665-MS.
Costamte, Y.R. 2010. Modelling of Open-Hole Gravel Packing on Longer Blank Pipe and Screen Sections. Master’s Thesis, Faculty of
Science and Technology, Universitetet i Stavanger, Stavanger, Norway, 15 June.
Fjaer E., Holt R. M., Horsrud P., Raaen A.M., and Risnes R. 2008. Petroleum Related Rock Mechanics, second edition. Elsevier.
Fjellstad, V., Statoil. 2007. Private communication
Forrest, J. 1990. Horizontal Gravel Packing Studies in a Full Scale Model Wellbore. Paper SPE 20681-MS presented at the 1990 Annual
Technical Conference and Exhibition, New Orleans, Louisiana, USA, 23-26 September. doi: 10.2118/20681-MS.
Gruesbeck, C., Salathiel, W.M., and Echols, E.E. 1979. Design of Gravel Packs in Deviated Wellbores. J Pet Technol 31 (1): 109-115; SPE
6805-PA. doi: 10.2118/6805-PA.
Harestad, K., Saasen, A. and Sterri, N. 1991. Fluid Mechanical Analysis for Control of Gravel Pack Jobs. Paper SPE 21672 presented at
the Production Operations Symposium, Oklahoma City, April 7-9. doi: 10.2118/21672-MS
Jain, A. K. 1976. Accurate Explicit Equation for Friction Factor. J. Hydl. Div. ASCE, NoHY5: May.
Jain, S., Chanpura, R., Barbedo, R., and Moura, M. 2008. Integrated Approach to Modelling Gravel Packs in Horizontal Wells. Paper SPE-
IPTC 12448-MS presented at the International Petroleum Technology Conference, Kuala Lumpur, Malaysia, 3-5 Dec. doi:
10.2523/12448-MS.
Larsen, O. 2010. Statoil Gravel Pack Test Sum-up. Slide SPE presentation presented at the 4th European Sand Management Forum,
Aberdeen, Scotland, UK, 10-11March http://www.spe-uk.org/Downloads/Sand-Management-Forum-2010/Presentations/
24%20Statoil%20Gravel%20Pack%20Tests%20Sum-up.pdf. Downloaded 24 December 2011.
Lyons W.C. 2009. Air and Gas Drilling Manual (Applications for Oil and Gas Recovery Wells and Geothermal Fluids Recovery Wells),
third edition. Elsevier Inc. Gulf Professional Publication.
Nguyen P.D., Fitzpatrick H. J., Woodbridge G. A. and Reidenbach V. G. 1992. Analysis of Gravel Packing Using 3-D Numerical
Simulation. Paper SPE 23792-MS presented at the SPE Formation Damage Symposium, Lafayette, LA, 26-27 Feb. doi:
10.2118/23792-MS.
Ojo, K.P., Osisanya, S.O., and Ayeni, K.B., 2006. 3D Numerical Simulator for Horizontal-Well Gravel Pack. Paper SPE 103291 presented
at the 2006 SPE Annual Technical Conference and Exhibition, San Antonio, Texas, USA, 24-27 Sept.
Oroskar, A.L., and Turian, R.M. 1980. The Critical Velocity in Pipeline Flow of Slurries. AIChE Jounal 26.
16 SPE 150546
Ott, W.K.P.E. and Wood, J.D. 2003. Modern Sandface Completion Practices Handbook, World Oil, second edition. Houston, Texas: Gulf
Publishing Company.
Pedroso, C., Sanches, E., Oliveira, N., Fernandes, L., Mickelburgh I.J., and Coelho, G. S. 2005. New Solution to Extend the Application
Window of Horizontal Open Hole Gravel Packing. Paper SPE 96257-MS presented at the 2005 SPE Annual Technical Conference and
Exhibition, Dallas, Texas, 9-12 October. doi: 10.2118/96257-MS.
Penberthy, W.L., Bickham, K.L., Nguyen, H.T., and Paulley, T.A. 1996. Gravel Placement in Horizontal Wells. Paper SPE 31147-PA
presented at the SPE International Symposium on Formation Damage Control, Lafayette, Louisiana, USA, 14-15 February. doi:
10.2118/31147-PA.
Penberthy, W.L.Jr., Echols, E.E. 1993. Gravel Placement in Wells. J Pet Technol 45 (7): 612-613, 670-674; SPE-22793-PA. doi:
10.2118/22793-PA.
Perry, R.H. and Chilton, C.H. 1973. Chemical Engineer’s Handbook, fifth edition. McGraw-Hill.
Sanders, M.W., et al. 2002. Gravel Pack Designs of Highly-Deviated Wells with an Alternative Flow-Path Concept. Paper SPE 73743-MS
presented at the SPE International Symposium and Exhibition on Formation Damage Control, Lafayette, LA, 20-21. doi:
10.2118/73743-MS.
Shah S.N. and Lord, D.L. 1990. Hydraulic Fracturing Slurry Transport in Horizontal Pipes. SPE DE, 5, 225-232. doi: 10.2118/18994-PA
White, F.M. 1991. Viscous Fluid Flow, second edition. McGraw-Hill Book Co.
SPE 150546 17
Case I: OD screen is larger than half of the wellbore diameter (DS > RW)
Derivation of the hydraulic diameter for a fully eccentric annulus configuration between open-hole and screen in a horizontal
section for this case was done by Chen (2008). This case is divided into a three segments calculation to compensate the
increasing of dune height.
Fig. A.1 – Geometry of case I and II where H<Rs, Rs<H<Ds, and H>Ds. Case II with H >Ds and Ds <H<Rw are described in this figure in
the lower side.
RS<H<DS: Dune height is less than the OD of the screen and larger than the outer radius of the screen (Fig. A.1).
(π − α ) Rw 2 + ( Rw − H ) Rw 2 − ( Rw − H ) 2 + ( β − π ) Rs 2 − ( Rs − H ) Rs 2 − ( Rs − H ) 2
Dh = 2.
(π − α ) Rw + (π − β ) Rs + Rw 2 − ( Rw − H ) 2 − Rs 2 − ( Rs − H ) 2
18 SPE 150546
(A-6)
H>DS: Dune height is larger than the OD of screen (Fig. A.1).
αRw 2 − ( H − Rw) Rw 2 − ( H − Rw) 2 (A-7)
Dh = 2.
αRw + Rw 2 − ( H − Rw) 2
Case II: OD screen is less than half of the wellbore diameter (DS < RW)
Four segments are determined in the calculation of hydraulic diameter to compensate the increasing of dune height in the case
where DS<RW. This is illustrated in Fig. A.1. These segments are equal to the Case I segments except for the introduction of a
new segment for DS<H<RW. In this case the hydraulic diameter is given as shown in Eq. A-8.
(π − α ) Rw 2 − ( Rw − H ) Rw 2 − ( Rw − H ) 2 (A-8)
D h = 2.
(π − α ) Rw + Rw 2 − ( Rw − H ) 2
Start
ρ f ,dp,µ f , g
Guess vs = 0.03
ρ f vs d p
Rep =
µf
4 ( s − 1).d p .g
v's = .
3 CD
v 's + vs
v's −vs < 0.0001 v 's =
2
by the fluid grows until it equals the weight of the particle. When the weight and the drag are in balance the particle no longer
accelerates but falls at its terminal settling velocity. The buoyancy and the drag force act vertically upward, whereas the
weight forces act downward. The settling velocity of a sphere in a fluid at rest is found following Eqs. A-12 and A-13.
FG = FD + Fb (A-12)
4 ( s − 1).d p .g (A-13)
vs = .
3 CD
The value of the drag coefficient depends upon the flow regime around the particle, which is the Reynolds particle number,
defined as shown in Eq. A-14.
ρ f vs d p (A-14)
R =
µf
ep
Correlation of Reynolds number and drag coefficient for rigid spherical bodies can be found for example in Fjaer et al.
(2008). This is a formula for calculating drag force in completely laminar environment, for perfectly spherical bodies. For
Rep<105 and typical for non-smooth spheres, the approximate expression is given by Eq. A-15.
24 (A-15)
CD ≈ 1 + 0.2 Rep + 0.0003Rep
2
Rep
Since most likely the Reynolds number will be in the range 1< Rep <105, a substitution of Eq. A-15 into Eq. A-13 will give a
formula for the settling velocity as shown in Eq. A-16.
1 Rep .( s − 1).d p .g (A-16)
vs =
2
.
18 1 + 0.2 Rep + 0.0003Rep
2
The calculation is solved by doing iteration procedure. It is started by guessing a value of vs (input) until it is matched with the
vs’ (output) as shown in Fig. A.2.
0.816
ρp − ρf Dh − ρ s
0.775
(A-23)
V1 = 0.0251gd p
ρ µ
f f
and
0.5
2 gDh ( ρ p − ρ f ) (A-24)
V2 = 1.35
ρf
Table A.1–Absolute roughness of pipe and open hole wall of rock formation for several conditions.
Absolute Roughness (Ɛ) Description
1.524 x 10-5 m or 0.0006 in New tubing
1.828 x 10-4 m or 0.0072 in Tubing and line pipe that has been in service for some time
2.286 x 10-4 m or 0.0090 in Very dirty pipe
0.003 to 0.006 m or 0.12 to 0.24 in Competent, low fracture
0.006 to 0.009 m or 0.24 to 0.36 in Competent, medium fracture
0.009 to 0.012 m or 0.36 to 0.48 in Poor competence, high fracture
Wall Roughness
The determination of the value to use for pipe wall roughness in the friction factor equations is sometimes difficult. The
parameter ε is not a property that is physically measured. Rather, it is the sand grain roughness that would result in the same
friction factor. Summary of absolute roughness from Beggs (2003) and Lyons (2009) are given in Table A.1.
SPE 150546 21