Professional Documents
Culture Documents
Abstract
The fracturing and gravel packing sand control technique,
most commonly referred to as FRACPACK, has recently
become one of the most successful and reliable ways to
complete low to moderate permeability unconsolidated
sandstone reservoirs. In addition to providing a good sand
control completion, FRACPACK offers the advantage of
stimulating the reservoir and reducing the final skin.
FRACPACK treatments are most commonly done with the
screen in place where a gravel pack packer is run with the
blank pipe and screens placed across the zone of interest.
Later, a fracturing treatment is performed through the gravel
pack packer and across the screens (Fig. 1)
Although a lot of research has been made in the fracturing
area of FRACPACKs, little is known about the nature of
slurry flow in the screen/casing annulus below the gravel pack
packer crossover ports and before entering the perforations. It
is believed that as pumping progresses, deposition of proppant
on the screen yields stable nodes that may cause bridging in
the annulus that may result in early termination of the
fracturing treatment.
This paper discusses the steps involved in mathematically
predicting node buildup on the screen during FRACPACK
treatments. A simple model was developed that calculates
the size of the node deposit at the end of fracturing treatments
based upon treatment parameters. FRACPACK case histories
from the Gulf of Mexico where wellbore screenouts occurred
were predicted by the model.
Introduction
Although the basic design of a FRACPACK job involves
fracturing technology, an important part of the design which
is often neglected is the flow of the fluid below the crossover
Mazen H. Omari
flow path, i.e. the flow into the perforations, the flow will
re-enter at the uppermost section again. This would redivert the flow back into the upper section of the screen and
deposition of the node should again progress from the top,
but now through a larger section of the screen. This
dynamic process continues with little resultant growth in
node length.
Due to node deposit buildup in the annulus, velocities in the
annulus of the casing/screen become high, causing greater
friction pressures and eventually more flow would be
diverted into the screen. This, coupled with the fact that
slurry at the final stages will be high in proppant loading,
could cause a very rapid build up of the node, eventually
plugging the annulus.
Basic Assumptions for Model
Even with the simplified flow pattern described above, it is
evident that finding exact equations to describe the flow
below the crossover ports would not be practical.
The distribution and redistribution process of the flow, as
well as the dynamic buildup of the node, make it difficult to
theoretically describe the process. Flow is non-uniform, even
in the screen/casing annulus, as it decreases from a maximum
value before the top perforations to zero at the bottom
perforations. To address this flow issue and estimate the
distribution as well as the node buildup process, a set of
assumptions were made.
1. The fluid flowing in the annulus of the screen/casing is a
slurry of uniform proppant concentration in all directions.
2. The slurry rheology is a direct function of the carrier fluid
rheology. A factor will be used that is a function of the
proppant concentration.
3. The fluid leaking off into the screen has the same
properties as the carrier fluid.
4. The node buildup across the screen will be assumed to
take place across a predetermined length of the screen to
be called node length. All the flow into the screen will
be assumed to take place in that section. Although this is
not strictly true, it will produce conservative results in the
model. (A value of 10% of the total screen length for the
node length was found to give the best match between the
model and field results.)
5. The equivalent path of flow will be defined as the length
of the screen where the total (equivalent) flow inside the
screen will exit into the perforations. This is an assumed
value and has no physical meaning, but is required to
calculate the pressure drops outside and inside the screen.
This value will be assumed to be 40% to 50% of the total
length of the screen.
6. The node buildup process across the screen is stable and,
once it develops, is not reversible. No erosion of the node
will take place, even when higher velocities are
encountered at a later stage of the treatment. This
assumption is true for very viscous carrier fluids.
7. The flow distribution between the annulus of the
screen/casing or outside and the annulus of washpipe/screen or inside is a function of the pressure drop
along the screen, pressure drop inside the screen and
SPE 37735
Pf
f
NRe(c)
13.476 x 4Q(1)
(D22- D12 )
= 2.28 x x V2-n' x (D2- D1)n' ....(2)
144n' K' (2n'+1)
3n
= 0.039 x fx LV2 .. (3)
0.8165(D2-D1 )
= 16 ..(4)
NRe
= 3470 - 1370n'(5)
Pf
Q
Nre
n'
K'
f
13.476 x 4Q ..(6)
(D22- D12 )
= 928 x x0.8165 V x (D2-D1 )n' (7)
CF = r (Laminar Flow)(11)
CF = r0.55r0.45 (Turbulent Flow)......(12)
r = ( 1 + [ 0.75(e1.5n-1)e-(1-n)/1000 ] ) (1.25 p )(13)
1- p
p = Cp /( Cp + )
Cp = Particle Concentration, lb m/gal
p = Particle Density, lb m/gal
Mazen H. Omari
Model Algorithm
The following are the steps involved in the calculation of the
output variables listed above:
a) From the carrier fluid density, absolute volume of
particles and proppant concentration, determine the
slurry density.
b) From Equations (11), (12) and (13) determine the slurry
friction correlation coefficient assuming laminar flow and
a shear rate of 100 sec-1. (The slurry friction correlation
coefficient is not very sensitive to the value of shear)
c) Assume a flow inside and outside as a function of the
total flow rate.
d) Calculate the velocity of flow outside and inside using
Eqs. (1) or (6). Use the density of slurry for outside flow
and density of carrier fluid for the inside flow.
e) Calculate the Reynold's numbers outside and inside from
Eqs (2) or (7) depending on the nature of the flow
(laminar or turbulent).
f) Calculate friction coefficient inside and out using Eqs (4)
or (9).
g) Calculate pressure drop outside and inside using Eqs (3)
or (8). Multiply the friction outside by the correlation
factor.
h) Calculate ' and from Eqs (15) and (16).
i) Calculate the pressure drop across the node using Eq.
(14). Note this is zero in the first stage of the treatment.
j) Check the equality in Eq. (17) by subtracting both sides.
k) According to the result obtained above, modify the flow
assumption and repeat until convergence or until the
equality in Eq. (17) is satisfied.
l) After calculating the flow inside or the flow into the
wash-pipe/screen annulus, calculate the sand depositing
on the screen by multiplying the flow rate by the
pumping time of the current stage and by the proppant
concentration.
m) Divide the value obtained in step (e) above by the surface
area and calculate the width of the node.
n) Repeat all the above steps for all the treatment stages
adding the node width cumulatively each time. Note that
as node builds up, so does the pressure drop across the
node as it is a direct function of the node width. Each
stage has a unique solution. The cross-sectional annular
area formed by the node has to be subtracted from total
area available for flow in the screen/casing annulus.
o) Screenout occurs if the node width becomes greater or
equal to the annulus of screen/casing.
At the end of the algorithm, all output variables were
displayed in a table form to check the development of flow in
each stage.
Results
In order to evaluate the model predictions, 23 jobs were
studied from a database of more than 200 fracturing jobs.
These jobs were chosen with varying casing sizes (i.e. 5 1/2,
7, 7 5/8 and 9 5/8 casing). Those jobs represent the
treatments where the highest amount of proppant was
designed and pumped.
SPE 37735
After inputting the job parameters and running the model for
all the treatments, it was found that using a node length of 0.1
times the screen length it gave the most reasonable results
when compared to field data (TABLE 1). The following
observations were made when comparing the actual
treatments results of treatments with the model predictions:
In the jobs where highest amounts of proppant were actually
pumped without screenout, the model predicted screenout
just before placing all the proppant. This was attributed to
the conservative assumptions that were used in the model.
From field results, it was clear that most jobs pumped with
a large amount of proppant were the jobs done using a
prepacked screen. The model supports this observation,
since the node width predicted in most cases was low
compared to that calculated for regular non-prepacked
screens.
Very few jobs were available in the database where a lot of
proppant was pumped in the case of non-prepacked screens.
The four cases studied show that the model predicted
screenout earlier than that encountered in the field. In case
there was no screenout predicted, the final node width
predicted by the model was high (50% to 70% of annular
clearance) indicating the likelihood of screenout had the
treatment continued.
The results of the model confirmed that some screenouts
were actually not caused by bridging due to node buildup.
This was also evident since such screenouts happened very
early in the treatment.
In general, the results of the model are conservative and
may be used to estimate the amount of proppant that can be
safely pumped prior to screenout.
The model predicted that a flow as much as 0.01 to 0 0.05
BPM in the screen/wash-pipe annulus is enough to cause
the node to buildup and eventually screenout.
Effect of Treatment Well Parameters on Screenout
A study was performed to determine the predicted effect of
pumping and well schematic parameters on the node buildup,
as predicted by the mathematical model. It appears that the
results of this study justify existing gravel packing operational
assumptions.
Wash-pipe Diameter
The wash-pipe to screen ID ratio has a pronounced effect on
the node buildup for 7 casing . In the case of a 9 5/8"
casing, and for the common size of screens used, the model
results suggest that wash-pipe ratio is not as significant as in
smaller size casings. This is due to the large annulus
available for the flow of slurry compared to inside of the
screen. (Fig. 4)
Variation of Proppant
Although most of the fracturing treatments were done using a
20/40 mesh sand, it is worthwhile to mention that the higher
the permeability of the proppant, the more likely a screenout
will occur. The situation is especially critical if 10/20 gravel
is used for fracturing. (Fig. 5)
Screen Diameter
The size of the screen was found to be a main factor in the
node buildup. Larger screen diameters lead to less annular
clearance which accelerates the screenout process. Some
Mazen H. Omari
Acknowledgments
The author wishes to thank Schlumberger Dowell for the
opportunity to prepare this paper. In addition, there are
numerous other people that have provided valuable input.
These include: Paul Massebouef, Ray Tibbles, Mehmet Parlar,
Norman Chesnut, Charlie Carroll, Michael Mathews and
Mike Colluzi.
DrillPipe
Screen
Casing
References
1. C.D. Haynes and K.E. Gray: Sand Particle Transport in
perforated Casing, SPE 4031
2. C. Gruesbeck and R.E. Collins: Particle Transport through
perforations, SPE 7006
3. Richard M. Hodge: Gravel Transport in Deviated Wellbores,
SPE 10654R.
4. S. Torrest: Deposit buildup During Gravel Packing with Viscous
Polymer Solutions, SPE 10072
5. Robert S. Torrest: Aspects of slurry and Particle Settling and
Placement for Viscous Gravel Packing ( AQUAPAC) , SPE
11009
6. Robert S. Torrest: The Flow of Viscous Polymer Solutions for
Gravel Packing Through Porous Media, SPE 11010
7. B.C. Skaggs: Transport Efficiency of High-Density Gravel
Packing Slurries, SPE 12480
8. Calculation of pressure losses across gravel packs, SPE
17167
9. D.G. Gurley and T.E. Hudson: Factors Affecting Gravel
Placement in Long Deviated Intervals, SPE 19400
10. P.H. Winterfeld and D.E. Schroeder Jr.: Numerical Simulation
of Gravel Packing, SPE 19753
11. M.H. Johnson J.P. Ashton and Hang Nguyen: The effects of
erosion velocity on filter-cake stability
During Gravel
Placement Openhole Horizontal Gravel pack Completions,
SPE 23773
12. Well Cementing by Eric B. Wilson.
Node Buildup
Slurry Flow
Carrier Fluid Flow
Wash
Pipe
Perforations
Figure 2 - Flow distribution and node buildup below the crossover ports
in a Frack and Pack
Total BPM :
Slurry
Gravel Pack
Packer
Screen
Node
Buildup
Lnode
Leqv
Qoutside : Slurry
DeltaP : Pre-pack
DrillPipe
Gravel Pack Packer
Crossover ports
Casing
Slurry Flow
Carrier Fluid
Wash
Pipe
Perforations
Screen
2
1.5
7 in
7-5/8"
9-5/8"
1
0.5
0
0.8
0.7
0.6
0.5
Washpipe Ratio
Figure 1 - Typical FRACPAC Schematic with screens in place and packer set.
Figure 4 - The node buildup increases as washpipe ratio decreases. The effect is less severe in 9 5/8 casing size
Values of nodewidth/annular clearance of greater than one indicate wellbore screenout
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
40-60
7 in
7-5/8"
9-5/8"
20-40
20-10
SPE 37735
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
7 in
7-5/8"
9-5/8"
10 BPM
Proppant Mesh
25 BPM
1.4
1.2
1
0.8
0.6
0.4
0.2
0
7 in
7-5/8"
9-5/8"
3-1/2"
4"
Figure 8 - Variation of the size of node buildup with crosslinked and linear HEC carrier fluids
15 BPM
Pump Rate
Figure 5 - The node buildup increases as the permeability of the proppant increases.
Values of Node width/annular clearance of greater than one indicate wellbore screenout
2-7/8"
2
1.5
5-1/2"
7- in
7-5/8"
1
0.5
0
50
100
150
Screen Length
Screen Diameter
Figure 9 - Effect of the length of interval (or screen length) on the node buildup as simulated by
the model.
Figure 6- The node buildup increases as the diameter of the screen increases.
3
2.5
2
1.5
1
0.5
0
X-linked
7 in
7-5/8"
9-5/8"
Linear
80lb
Linear
60 lb
Carrier Fluid
Figure 7 - Variation of the size of node buildup with crosslinked and linear HEC carrier fluids
Casing
Csg Weight
Casing ID
Screen
Screen Type
Screen ID
Screen OD
Scr. Gauge
Scr. Length
Wash-pipe
OD
Prop. Mesh
Carrier Fluid
BHST
Frac BPM
No. Stages
Design lbs
Actual lbs
Case 1
Case 2
Case 3
Case 4
Case 5
5 1/2
17 lb/ft
4.892
2 7/8
Prepak
2.441
3.42
0.008
153 ft
1.90
7
29 lb/ft
6.184
3.5
Prepak
2.992
4.045
0.008
360 ft
2.37 5
7-5/8
33 lb/ft
6.76
4.5
Prepak
3.94
5.2
0.008
150 ft
2.875
9-5/8
47 lb/ft
8.681
4.5
Regular
4.045
5.196
0.012
130 ft
2.875
7-5/8
29 lb/ft
6.875
4.0
Regular
3.5
4.54
0.012
100 ft
2.875
20/40
HEC80
130 F
12
6
125000
48 000
20/40
HEC80
180 F
20
6
190000
175000
20/40
HEC80
180 F
13
8
196000
147000
20/40
HPG30
170 F
25
11
95 000
74 000
20/40
HPG35
186 F
12
7
70 000
57 000
76 000
------
180 000
------
115 000
------
95 000
0.50
54 000
------
Model Predictions
lbs Proppant
Node size
Five critical cases out of the 24 cases used to validate the model. The
results predicted were close to what was observed in the field . Cases 1
and 3 are examples of situations where wellbore screenout was not
accounted for in the design stage. In case 2, results of field treatment
and model show that the amount of proppant planned was really the
optimum for that particular schematic and pumping schedule. Cases 4
and 5 demonstrate the advantage of using prepacked screens.