Вы находитесь на странице: 1из 17

To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323.

Copyright © 1997 Elsevier Science Ltd

Copyright © 1997 Elsevier Science Ltd


Int. J. Rock Mech. & Min. Sci. Vol. 34, No. 3-4, 1997 ISSN 0148-9062
To cite this paper: Int. J. RockMech. &Min. Sci. 34:3-4, Paper No. 323

P H Y S I C A L M O D E L I N G OF SAND P R O D U C T I O N
H . H . V a z i r i l; R . P h i l l i p s 2 ; S. H u r l e y 2

1 Department of Civil Engineering, Technical University of Nova Scotia, Halifax, Nova Scotia, B3J 2X4
Canada
z C-CORE, Memorial University of Newfoundland, St. John's, Newfoundland, A1B 3X5 Canada

ABSTRACT
Centrifuge experiments were conducted to identify mode of failure following sand production (e.g.,
erosional channels or enlarged cavity), and to quantify impact of sand production on flowrate. Fifteen
tests were performed all indicating that sand production resulted in an enlarged cavity that was
cone-shaped. The cone extended to the top surface of the sand with a slope angle of about 35 °. Typically,
the improvement in the steady state flowrate was ten fold. These results indicate that sand production can
provide a cost effective means of boosting production by altering the permeability characteristics of the
formation and the flow regime (from a predominantly spherical to radial flow situation). The centrifuge
tests neither captured the conditions resulting in formation of long erosional channels nor the several
orders of magnitude improvement in production as inferred from some field projects in the Clearwater oil
sand formation. A comprehensive discussion is presented to tie the physical model test results with the
anomalous field data.
Copyright © 1997 Elsevier Science Ltd

KEYWORDS

w e l l b o r e c a v i t a t i o n • s a n d p r o d u c t i o n • e r o s i o n a l c h a n n e l s • c e n t r i f u g e test • oil s a n d •
c o n s o l i d a t i o n • f l o w i m p r o v e m e n t • plastic failure

INTRODUCTION
Drilling of wellbores, or opening of a cavity, results in shear failure of the formation around the cavity.
The radius of shear-failed zone is largely governed by the strength properties of the formation and
increases inversely with strength. Unless the formation is unusually weak/friable/brittle, the drilling
process will not result in sloughing of the wellface material or cavitation. Potential for cavitation
(enlargement of the original drill opening) and solids production increases with the rate of seepage. In
general, most oil and gas operating companies attempt to minimize or eliminate the production of solids
which causes delays and direct costs to cleanup the affected wells. Some of the significant benefits of
wellbore cavitation and solids production, in general, have not been well understood or adequately
exploited. A clear demonstration of the positive impacts of solids production on the performance of oil
and gas reservoirs forms the thrust of this paper.
There are a number of field cases that show improvement in fluid production resulting from solids
production. In the Fairway region of the San Juan Basin (Colorado/New Mexico), it has been noted that
gas production from coalbed methane reservoirs can be improved by several fold if openhole cavity

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

completion technique, which encourages wellbore cavitation, is adopted (Palmer et al. 1995). In oil sand
deposits of Alberta (Canada), the field cases that have shown a significant boost in bitumen production
due to sand production include Clearwater (Yeung 1995) and Lloydminster regions (Loughead,
Saltuklaroglu 1992).
In coalbed methane, there are clear indications that much of the coal produced comes from wellbore
cavitation; these have been confirmed by gamma ray logs which show the entire profile of the cavitated
zone. For instance, at the COAL Site (Palmer et al. 1995), it was observed that the wellbore radius
increased from 0.15 m to about 1.4 m within the 10-m thick coal formation. In oil sand formations,
however, there is no general agreement as to the mode of failure. Sand production can be due to
cavitation or due to development of wormholes (channels of high porosity caused by erosion of sand) or
perhaps a combination of these. A knowledge of the mode of failure is essential for understanding the
true mechanisms resulting in improved performance in reservoirs that undergo sand production. To
provide an answer to this question, centrifuge experiments were performed to identify the mode of failure
following sand production, and to quantify the impact of sand production on flowrate. The relevant
results and their discussion are presented here preceded by a brief background of the problem and the
previous work reported on this subject.

P O T E N T I A L M E C H A N I S M S R E S U L T I N G IN F L O W I M P R O V E M E N T
In a number of fields in various regions of Alberta, such as Lloydminster and Clearwater, it has been
reported that in wellbores experiencing sand production, the subsequent recovery rate was appreciably
improved. To tie in the improved production with sand production, several different theories have been
postulated including: (1) increase in sand porosity due to shearing and stress redistribution; (2) solution
gas drive; (3) creation of higher conductivity channels or wormholes; and (4) non-Newtonian fluid flow.
The proceeding provides a brief description of these mechanisms.
Vaziri et al. 1990, 1997a, 1997b have shown that following production of the solids, radius of the well
within the affected area becomes enlarged. Enlargement of the cavity changes the stress state around the
wellbore over a considerable distance (approximately 3 to 10 times the cavity radius) depending on the
formation strength properties. The stress change within this disturbed zone results in a net increase in
porosity and hence permeability due to: (1) shear-induced dilation, and (2) a net reduction in the mean
effective stress. Numerical studies that follow these physics have shown good agreement between the
measured and computed wellbore size and production increase in coalbed methane reservoirs.
Solution gas drive is generally considered to be the main drive mechanism (Smith 1988). The formation
of bubbles in the oil is believed to maintain the pressure in the reservoir for a longer time (Maini et al.
1993). The predictions based on this theory were in agreement with the best performance achieved at
Lloydminster without recourse to negative skin or other permeability enhancement beyond laboratory
values. In many cases solution gas drive by itself is not sufficient to explain the enhanced oil recovery.
Significant increases in oil production occurred only when large quantities of sand were produced.
Experiments in sand packs, with no sand production, have shown that the permeability of the pack to oil
is not increased when gas bubbles are generated (Tremblay et al. 1996).
Based on injectivity tests conducted after six months of primary production at two Mobil Celtic pilot
wells, Loughead, Saltuklarogla 1992 suggested the existence of high permeability channels in the
reservoir. Numerical simulation of the pressure build-up assuming linear flow into induced high
permeability channels was found to be consistent with the measured data. Tracer experiments between

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

injection and production wells have been performed in the field by several operators to measure the
tracer travel between wells after considerable sand production took place (Squires 1993). In general, the
travel time was at least an order of magnitude shorter than that normally predicted for unaltered
formations. Such short communication times can be explained by the formation of wormholes (Tremblay
et al. 1996).

Poon, Kisman 1991, using a radial non-Newtonian diffusivity equation, found that dilatant
non-Newtonian fluid behaviour could be a significant mechanism in the primary production of heavy oil.
The behaviour can be beneficial in the reservoir where shear rates are low and can help explain how high
production rates can be sustained over an extended period of time.

ESSENTIALS OF A FIELD P R O J E C T INVOLVING SAND P R O D U C T I O N

Background
Yeung 1995 described a field project, referred to as the Burnt Lake project, involving sand production
with a concomitant improvement in primary (cold) production. The site is located 300 km Northeast of
Edmonton in the Cold Lake region. Suncor and several other partners (e.g., Amoco) were the operators
of this project. In total, 44 wells on two clusters and 7 delineation wells had been drilled and cased.
Suncor began experimenting with primary production on four of these wells. It was noted that with
improvements in progressive cavity pump design, primary production had greatly increased in
Clearwater heavy oil reservoirs in the Lindenberg/Elk Point area. Typically the production rate of these
wells were about 20 m3/day; peak bitumen rates of 40 m3/day were achieved with larger progressive
cavity pumps and the injection of viscosity reducing chemicals down the annulus. Sand production was
quite high at the beginning but it tapered off to very little within the first year. During the operation, sand
production rate would increase with an increase in the pump speed or production rate; once the
production rate stablized, sand production would cease.
Encouraged by the success of the above well tests, a major project was undertaken involving 17 wells. In
general, oil production level was inconsistent and field operations ran into difficulties. Production in
some cases was about 5 to 10 m3/d but most of the wells encountered problems due to localized gas cap,
production of high volumes of gas, and water production. Based on gamma-ray logs, it was considered
that water production was caused by the deterioration of the shale cap which provided access to the
upperlying aquifer.

Field Tests
Using the standard single-phase radial Darcy flow, the bitumen production rates at Burnt Lake should
have been about 0.1 m3/day per well. However, sustained production in excess of 20 m3/day was
recorded at three wells that had experienced sand production. It was speculated that sand production
resulted in development of wormholes which provided the conduits for higher production (Yeung 1995).
To validate this hypothesis, the following field experiments were conducted.
Dye tests showed that communication readily occurred from those wells with shale cap damage to the
newer wells which had no indication of shale damage but nevertheless watered out (it is presumed that
water entering the oil sand layer from the overlying acquifer above the damaged shale cap has been able
to flow through a set of wormholes to the newer wells). After dye was injected, it showed up in a matter
of few hours in an observation well about 500 m away.

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

A seismic program was also undertaken to detect changes in density; detection requires an appreciable
change in porosity (say greater than 15%) over a thickness of at least 6 m (Yeung 1995). In the three
lines of survey performed that covered a 6 km length, anomalous zones were identified. Along the line
where the total sand production was in excess of 1000 m 3 (1 St line) change in the seismic response was
detected. However, over another section where more than 500 m 3 of sand were produced, no change in
seismic response was detected. Along the third line anomalous results were detected which were not as
profound as those encountered along the 1St line, but these anomalies are believed to be due to gas
accumulation as the sand production was minimal (Yeung 1995).

Laboratory Simulation of Wormholes


Tremblay et al. 1996 performed a set of laboratory experiments to simulate development of wormholes.
In these experiments, heavy oil (viscosity = 27,000 cp) was injected through a horizontal sand pack into
an orifice simulating a perforation (6 m m in diameter). The sand was Clearwater oil sand that was
densely packed to a porosity of about 35%. The length of the pack was 0.36 m, its diameter = 0.1 m; the
injection pressure was about 2 MPa. Triaxial tests performed on samples of this pack indicated a friction
angle of 50 ° .
The flowing oil induced the production of sand from the pack when the pressure gradient at the orifice
reached a certain critical value. The sand pack was scanned and the images revealed that a high
permeability circular channel (approximate diameter = 40 mm) had formed in the pack while sand was
being produced. The porosity within the wormhole was much higher (about 55% and in some sections
100%) than the porosity of the undisturbed sand pack. No appreciable change in porosity was observed
in the vicinity around the wormhole. Based on the linear flow experiments performed, it was concluded
that channels of higher porosity can occur in dense, heavy oil sands. Other tests involving water instead
of bitumen or sand that was not densified did not materialize in wormholes (the properties were not
conducive to stable formation of wormholes). In those tests, sand production typically resulted in
formation of a cavity at the top of the pack after the wormhole collapsed.

CENTRIFUGE TESTS

Test Setup
To study the mode of sand production and its impact on flowrate, a set of centrifuge model tests was
performed. Schematic details of the test set up are shown in Figure. 1. The setup involved an 20-mm
diameter casing placed centrally within a 900 m m diameter cylindrical box. The outflow was through a
set of three 3-mm wide perforations, referred to a slits on Figure. 1; each arc length of each slit is about 5
mm.
The liquid head was controlled by adjusting the elevation of an external overflow using a DC motor. For
each test, the initial stage allowed seepage with no sand production to monitor the flowrate for a given
head difference. This was accomplished by placing a cylindrical porous stone at the perforations level.
To permit sand production, a vertical actuator was used to lift up the porous plug. The outflow fluid was
recirculated back to the chamber by a pumping system which was made up of two cylinders; one was
used for driving and the other for pumping. The sand transported through the perforations was collected
in a sand trap under the casing. Three load cells were placed at the base of the sand trap to record the rate
of sand production. The pore fluid pressure in the sand layer was monitored by six miniature pore
pressure transducers (PPT); a seventh PPT placed in the sand trap provided an independent check of the

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

fluid head in the well.


Some of the details of model construction are as follows. The soil used in the experiment was fine,
blasted silica sand, type # 00. Dry sand was rained through a series of sieves from a height of
approximately 800 mm; this depositional process densifies the sand to a relative density in the range of
75 % to 85 % (porosity of about 41%). The sand layer was underlain by an impermeable plastic sheet to
eliminate under drainage. Drained and undrained triaxial compression tests performed on samples of the
same fine sand as used in the centrifuge test and prepared at 60 to 65% relative density showed an angle
of friction of 35 °. A very strong dilatant response was observed during shearing which in the undrained
tests manifested itself by generation of high negative excess pore pressure.
To better identify the pattern of sand failure four blue sand circles were placed with an approximate
radius of 55, 110, 190 and 288 mm. Figure. 2 shows the final surface of the sand, marked by concentric
circles of blue sand, prior to being overlain by an impermeable geomesh system comprised of a
geomembrane and geofabric. Thickness of the sand layer above the slit was about 120 mm. Overburden
material, comprised of coarse sand and gravel, was placed over the geomesh to the top of the tub (about
80 m m in thickness). The placement of impermeable plastic sheets below and above the sand layer
ensured radial flow through the sand towards the opening in the well.

Specifics of Tests Performed


In total 15 centrifuge and three 1-g tests were performed. The variables considered in these tests
included: fluid saturation and viscosity, sand density, flowrate and centrifugal acceleration. In ten of
these tests involving homogeneous sand, the mode of failure was an enlarged cavity that was cone shaped
with no indications of wormhole initiation. In the subsequent 5 tests, special efforts were made to
encourage and facilitate wormhole development. In such tests, loose channels of sand (about 5 m m thick,
30 m m wide and 250 m m in length) were created in three directions, 120 ° apart, around the slits as
shown in Figure. 3. To facilitate stability of potential wormholes, a piece of thin plexiglass was placed
over each loose channel in one of these tests. Due to space limitations, the results of only two tests are
discussed here. Both of these tests incorporated loose sand channels, the difference between them was
that in one test (Test 15) the loose sand channels were not covered while in the other (Test 14) the
channels were covered with a thin plexiglass.
For both tests the following conditions prevailed. The sand thickness placed in the tub was 123 m m and
its relative density was 82%. The sand was initially saturated with water under vacuum which was then
displaced by Canola oil having a viscosity of about 50 cp (50 x 10-6 kPa.s), the fluid that was used for the
rest of the experiment. The oil and water interface resulted in adhesion between sand particles which is
estimated to provide an equivalent cohesion of about 5 kPa. The centrifugal acceleration was 24 g at the
slit level. The applied head difference between the well and the outer boundary of the tub was maintained
at around 6.0 m m (there were some slight fluctuations in the head difference during the course of each
test but these were considered to be inconsequential).
The test process involved first measuring the flowrate under steady state conditions while preventing
sand production. This data provides an important reference point for quantifying the influence of sand
production on flowrate. For both tests with the porous stone in place, flow was continued until steady
state condition was reached. Prior to removing the porous stone, the outflow leveller was lifted to its
initial position to resume a no flow condition. Then the porous stone was lifted out and the conditions
were monitored for several minutes. This was done to test if the sand remained stable under no flow

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

conditions (i.e., no free flow of sand through the slits due to gravity). In fact, the sand trap load
transducer readings confirmed that the removal of the porous stone did not culminate in any sand flux.
Next the outflow leveller was lowered to its previous flow position and that state was continued until no
further sand production or a change in flowrate could be observed. The results for each test are as follows.

Experimental Results
Test 15
For Test 15, under a head difference of about 6 mm, flowrate under steady state conditions before
removal of the porous stone was 2.15 ml/min. As shown in Figure. 4, the porous stone was removed at T
= 45 min; the head difference was then increased to 6.0 m m which resulted in an immediate production
of sand and a concomitant and dramatic increase in flowrate. Sand production continued until T = 150
min. The rate was 15.5 gr/min between T = 45 and 90 min and thereafter dropped quite sharply; between
T = 90 and 120 min the rate was 3.8 gr/min. The steady state flowrate at the end of the test was 24
ml/min which is a remarkable 11 times higher than that before the sand production under the same head
difference.
Figure. 5 shows the variation of pore pressure, as recorded by the six transducers placed within the sand
layer, at some selected times during the test. It can be observed from the T = 25 min profile,
corresponding to a steady state flow condition under a head difference of 6 mm, that before
commencement of sand production pore pressure is relatively uniform throughout but becomes very
steep within a close proximity of the well (40 m m from the well). This pore pressure profile is
characteristic of flow towards an orifice (spherical flow). Pore pressure profiles corresponding to
conditions after sand production reveal a pattern consistent with the development of an enlarged cavity.
At the end of sand production (T = 165 min), pressure profile becomes flat over the first 60 m m which
reflects the opening and the significant increase in permeability around the opening. Beyond this region
the profile typifies that forming around an open well (radial flow). The ten-fold increase in flowrate
shown in Figure. 4 is considered to be due to the formation of this enlarged cavity that has significantly
improved the permeability over an appreciable distance around the initial opening.
Finally, after completion of the test, the overburden sand and gravel was removed along with the
geomesh; dimensions of the cavity were recorded. Then the sand layer itself was carefully excavated by
working in from the tub boundary towards the well. Special attention was paid to capturing any potential
erosional channels as well as preserving the geometry of the enlarged cavity. Excavation of the
overburden material at the end of the test revealed formation of a cone-shaped cavity within the sand
layer. Radius of the enlarged cavity was 168 m m at the top and 39 m m at the base (slit level). Thickness
of the sand layer between the slit and top of the cavity was 68 mm. Close inspection of the cavity face
and the sand around the cavity did not indicate formation of erosional channels. It is worth pointing out
that at 24 g acceleration, erosional channels with an effective diameter of 30 m m could have been formed
while being visible (diameter of just over 1 mm) and stable in the post-inspection phase. Tremblay et al.
1996 experiments show that erosional channels with a diameter of greater than 30 m m do remain stable.

Test 14
This test was performed in a similar manner as Test 15 with the exception that the loose sand channels
were supported by a thin plexiglass to permit stability of the channels should they become open as a
result of sand erosion. The objective was to study the response under conditions that would favour
channelling and to test whether diversion towards channelling would eliminate or lessen the tendency for

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

cavitation. Figure. 6 shows the sand and flowrate variation. It can be observed that before sand
production, the steady state flowrate was 2.3 ml/min. After the porous stone was removed and sand
production commenced, flowrate sharply increased to about 65 ml/min followed by a steady decline to a
final level of 51 ml/min. Comparing steady state response before and after sand production indicates a
22-fold improvement in flowrate. Sand production rate in this test was steeper than Test 15 and was
completed over a shorter period. Production rate was about 26 gr/min between T = 110 and 145 rain
followed by a sharp termination after T = 150 min.
Figure. 7 shows a plan view of the enlarged cavity following excavation of the overburden material and
removal of the geomesh; the photo shows suspension of one piezometer within the enlarged cavity. Full
excavation of the face revealed the sand thickness around the cavity reduced to 65 mm and the maximum
radius of the cavity was 200 mm. Partially open drainage paths were observed under each plexiglass
sheet. Erosion of the sand was observed to start at the end of the plexiglass near the tub wall and progress
towards the opening as shown in Figure. 8. The loose sand originally poured within the channels was
mixed with coloured sand to better identify the transport pattern of the solids. Figure. 8 shows clearly
that the loose sand within the plexiglass-covered trenches had undergone erosion which was not the case
for Test 15 where the sand channels had been left uncovered. From comparison of Test 15 and Test 14
results in it considered that the flowrate difference can be largely attributed to the influence of the
erosional channels. Without the channelling, flowrate was improved by a factor of about 11 while with
the higher porosity channels the flowrate improved by a factor of 22. It is important to note that some
variations inevitably exist in preparing models and running the experiments which will influence the
results. It is, however, our contention that such variations do not change the response nearly enough to
account for doubling the flow performance from 11 times to 22 times the initial rate. Crude calculations
performed to estimate flowrate through porous pipes, similar to ones developed in Test 14, show that
flow can be improved by several fold in their presence.

DISCUSSION
The objective of this discussion is to compare the findings from our experiments with the observations in
the field and other laboratory studies in an effort to establish a better insight into the mechanisms that
result in improved production under primary conditions. The most relevant points, in a concise form, are
as follows.

Centrifuge Tests
The centrifuge tests performed were not intended to match the precise conditions in the field. The main
departures include: (1) the sand used did not have all the characteristics of in situ oil sand, in particular, it
was not as dense; (2) the fluid used was much lighter than bitumen (50 cp vs 50,000 cp); (3) the fluid
used did not include any dissolved gas; (4) natural heterogeneities and anisotropy, both in terms of
material properties and stress conditions, were not modelled; (5) stress and pressure conditions used were
much smaller than those in the field. Relevance of the following findings to the field must be considered
with these factors in mind.
1. The radial flow centrifuge tests show that sand production creates an enlarged cone-shaped cavity
with no evidence of wormholes. Improvement in flowrate is about ten times which can be attributed
to: (a) increase in the opening size, (b) change from spherical flow to radial flow as the cavity grows
to the top of the sand layer, and (c) increase in permeability within the plastic zone around the cavity
due to dilation and a net decrease in the mean effective stress level.

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

. An appreciable driving component for sand production observed in our experiments is the gravity.
This is manifested by a conical-shaped cavity right up to the cap; had gravity not been dominant,
zone of depletion would travel radially out in the plane of the slit resulting in channelling
(wormholes). In fact, as was seen in Test 14 that involved supported sand channels, the gravitational
effect still dominated resulting in a cone-shaped cavity.
. Although not presented here, numerical analysis of the problem has shown that radius of the plastic
zone is typically four times the cavity radius (the denser the sand the smaller is the plastic radius).
Level of permeability improvement within the plastic zone depends on the assumed
perm-porosity-stress relationship. By adopting a relatively strong, yet credible, perm-porosity
relationship, it can be shown that the influence of enlarged cavity and its associated plasticity is to
improve the flowrate by about a factor of three. Details of the numerical model used and its
application to simulation of openhole cavity completion in coalbed methane can be found in Vaziri
et al. 1997a.

Analysis o f Field Observations a n d Interpretations


Background and details of the field tests performed to establish potential geomechanical mechanisms
resulting in improved production, including a seismic program and dye injection test, were described
earlier. Below we examine the ramifications of the tests performed and the observations made.
1. The seismic tests suggest that along the 1st line of survey, wormholes must have coalesced to
culminate in a thick layer (> 6 m) of higher porosity material that extends beyond 500 m (this is the
distance between the observation and injection wells when the dye test was performed). Using
established geotechnical concepts that govern the mechanical response of sands, it is not possible to
extend the stress effects associated with sand production around an opening of 0.1 m radius to
distances further than 60 m from the well. This can be supported using finite element modelling
comprising various stress-strain and failure relationships as shown in Vaziri 1995. Such theories do
not support significant levels of sand depletion (or porosity increase) at +500 m as inferred from the
field data. It is possible that the anomaly in the field is partly due to gas accumulation.
2. The communication established between the injection and observation well through the dye test
suggests that either (a) the injection resulted in fracturing of the formation, or (b) the sand-depleted
zone was pancake shaped covering a distance in excess of 500 m, or (c) the wells happened to be
favourably aligned with existing conduits or fractures (this is somewhat unlikely). Had the sand
depleted zone been quite thick (i.e., thicker than 6 m), significant volumes of dye would have been
required to make the +500 m communication. The most natural location for the pancake is likely to
be just below the shale caprock.
3. The variability in production performance and the seismic data make it difficult to ascertain the true
conditions in the field. For instance, should sand production result in 200 times improvement in cold
production as was seen in one case or in more modest levels of improvement (e.g., ten times or less)
as were seen in other cases? Will sand production result in a seam of high porosity sand (pancake
type), or open circular channels created by sand erosion, or a thick layer of sand that undergoes a
uniform and appreciable increase in porosity? In fact, could sand production effects be more
localized resulting in an enlarged cavity (which may actually collapse during the well operation) and
a thick annulus of sand (say 10 m wide) around it of a significantly higher porosity due to large
strain, shearing and sand depletion?

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Field Relevance of Laboratory Simulated Wormholes


Following up on the description provided earlier on the experiments performed by Tremblay et al. 1995,
we consider the features that may not be representative of the field conditions include: (1) linear flow
(favouring a linear form of failure); (2) high pressure gradient (2 MPa over 0.3 m); (3) absence of
gravitational effects that otherwise would favour a conical-shaped cavity failure; (4) constraints on the
stress redistribution arising from the top rigid boundary; (5) reduction or relaxation of total vertical stress
resulting from sand production (in the field, the total vertical stress generally remains constant); and (6)
boundary conditions that preclude formation of an enlarged cavity emanating from the orifice (cavity can
only form at the top of the pack).

Plausible Explanations o f Field Observed Flowrates


The observed variability in the production performance in the field as described by Yeung 1995 does not
provide us with a reliable reference to validate our models and experiments. It is, however, interesting to
speculate on plausible scenarios culminating in a dramatic boost in production (e.g., greater than 20
times). Known mechanisms leading to flow improvement include development of erosional channels,
cavitation and its associated zone of plasticity, solution gas drive, and opening of natural fractures
triggered by stress changes arising from drilling of the wellbore and sand production.
Using numerical modelling, it can be shown that permeability increase in the plastic zone can improve
the flowrate by as much as 300%. Transformation of spherical flow around a single perforation (as was
adopted in our study) to a radial flow as a full cavity is formed can increase the flowrate by another
factor of three. This is basically the condition modelled in the centrifuge tests resulting in about a one
order of magnitude boost in flowrate. Although not experimentally modelled or numerically analyzed, it
is estimated that solution gas drive can increase the flowrate by a factor of three if the reservoir pressure
is brought down to about one third of its original value. As it can be seen from this crude analysis, it is
possible to postulate a scenario that can boost production by a factor of thirty through cavitation and
solution gas drive.
Another viable scenario that can account for both production enhancement and the rapid travel of dye
over a 500 m distance (assuming no fracturing) is as follows. Initially, an enlarged cavity was formed to
the top of the sand layer (beneath the shale cap) as spherical/cylindrical cavities forming around the
perforations coalesced. The strongest seepage path then started to form at the shale/oil sand interface. As
the sand lens closest to the shale interface is in the region of the largest cavitation (due to the cone shaped
nature of the cavity), it undergoes the greatest extent of yielding and hence the highest level of
permeability enhancement. (Note that the sand is particularly susceptible to erosion as it has yielded and
possibly strain-softened.) The diversion of stronger flow through this zone results in the erosion of the
sand. As the erosional process in the top sand lens continues, greater quantity of flow will be diverted
through it thus facilitating its further extension. By this stage, it can be assumed that the critical flowrate
has been reached. Critical flowrate is the rate that is sufficient to mobilize the flow of the sand particles.
This pancake-shaped cavity grows as long as the source can provide the required volume of fluid. It can
be noted that cavity beneath the shale cap can also occur as a result of collapse of wormholes. Therefore,
whether an enlarged cavity is initially formed or wormholes, erosion below the shale cap remains a
viable scenario.

SUMMARY AND CONCLUSIONS


In Canada, there are vast reserves of methane-rich coalbeds in Alberta and Maritime provinces (estimated

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

500 trillion m 3) and an abundance of bitumen-rich oil sands formations (estimated 500 Billion barrels) in
Alberta and Saskatchewan. In coalbeds and oil sand formations with favourable strength characteristics,
production can be improved significantly (e.g., by up to ten fold) if solids production is induced. Such
performance improvements in coalbed methane have been validated in certain fields of the San Juan
Basin where the openhole cavity completion technique was employed. In oil sands the published field
data are too sparse to warrant a reliable evaluation of the effectiveness of sand production on flowrate in
a consistent manner. There are, however, several compelling evidence of dramatic improvements that can
be realized as a result of sand production. Such field cases include Cold Lake and Lloydminster oil sand
formations. Our numerical and experimental studies support the effectiveness of solids production on
improving flowrate.
Based on a comprehensive experimental program involving fifteen centrifuge tests it was observed that
sand production resulted in an enlarged cavity (cone-shaped) with no indication of erosional channels.
Typically, the average cavity radius was ten times the initial well radius and the improvement in the
steady state flowrate was ten-fold. In one test where conditions permitted formation of stable erosional
channels, flowrate was improved by a factor of twenty. These results indicate that sand production can
provide a cost effective means of boosting production by altering the permeability characteristics of the
formation and the flow regime (transforming from a spherical flow to a radial flow).

ACKNOWLEDGEMENTS
The funding for this study was provided by NSERC and the Alberta Department of Energy. We
appreciate the assistance of Dr. A. Heidari in analysing the test data used in this paper.

FIGURES

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Paper 323, Figure 1.

De'
1:
Besm
I

1P|eg 1~" E I l ' b ~ - r e

-.-.-..
r
Gr+yel ~.

'•
i

Ph,mt:lt Nbet|s
i
• PPTI
/

Lo?.d ...I~# Ill- . ./~..


I "~ . . . . . . . . . . . . . . . --~

Y
::::::::::::::::::::
.. ................ [ C*ble

Figure 1. Schematic representation of the test setup

Paper 323, Figure 2.

I • . - . . .....
. i¸

" I

J
.~"

-. -. . . . . . . . . ~ .--"

Figure 2. Photograph of the final surface of the sand marked by the concentric circles

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Paper 323 Figure 3

~.o."~'.;:::.. o<.'"
• .. OO0 °

"'~''.... ~.'"
&

"-,L"~.'i ' "" "~~ : •~•,-.-,:~'


= ~'
~t I~% i~£i • %~<%:

,":.
..
.%.,...
.
• ~.
.....-'-2,
. ~÷
. ..~.,~.-

.;~,......... • ~,:,,~;'"

:: "b

Figure 3. Photo showing 3 uncovered loose channels of sand, 120 ° apart, around the slit

Paper 323 Figure 4

--'---'--'--'--=-" ' ........... -I "~

35

3r]

2~ ----

Pltl$ rcmo~'¢~ tr" •


N

Fl~wra~
a,C~? N
_="
P
0
0
P
ff ]9
or

o|

0
#

. . . . ! . . . . t . . . .

"rime (rm ~l

Figure 4. Flowrate and sand production response for Test 15

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

P a p e r 3 2 3 , F i g u r e 5.

' ' ' ' I I " ' ' : . . . . I ' . . . .

1 = ;~ ~n , . ~ . ~ ~
~ - ~ ° ..-;~

- "[ = G~ n-dn / ,,
0~ ~ ,Eft" ,,..,. ,..,
._,~..,-"

~. 7 "r='lOSmir.~ " ' t


F
/ "- "" " d 1~g rvJir
I ~.,,,~'/ '~-~" " " " ""
0 Ip~ ' -
/ I '',~ 0 T=15

Jl.i . , " . ,' T.,3Omin C -T-f,.5


--~- -T=~
"~'
[,t
- - X - - T - ',1~,~

//.~b~~ •-*.. T= :'~,0

~." I . . . . I I . . . .

O i~ 200 t00 4,)'j 5C~


Distance From Welt (rrm0

Figure 5. Pore pressure profile at various stages of experiment for Test 15

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Paper 323, Figure 6.

' " ' i ' " - , ~ . . . . . , . . . . . . . . j 70


. . . . . . m . . . . . . i ~ ~ ~ ~ l i ~.

t " "' I
8 (~3
[
[
[ I %, ~
t 5"
'~fJ i- tip ]~]0vcrzt: .~ ,~r)
l" Piu~i T=r.~o'.'~"J.
i

4(~.3 -
i
F
..
*
i
t ' 1 30
~,

20
d
d
- C~' -
l -2.
I0
I I
1

f_, ; ..... :-i ......... ,........ 22...7; ' 2 i ',2 222; Z.2:._.:.iL ; t
.~,~. I Of J 151:. Z(:0 250 311(I ]50

I-in~e Cmiei

Figure 6. Flowrate and sand production response for Test 14

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Mm. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Paper 323, Figure 7.

Figure 7. Photograph after excavation showing the enlarged cavity and a suspended piezometer

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

Paper 323, Figure 8.

~-.. -, ..

.'--..

.,J

Figure 8. Photo of erodded plexiglass covered channels near the well

References

References
Loughead D.J., Saltuklaroglu M. 1992. Lloydminster Heavy Oil Production: Why so unusual? 9th Annual
Heavy Oil and Oil Sands Technology Symposium, Calgary.
Maini B.B., Sarma H.K., George A.E. 1993. Significance of foam-oil behaviour in primary production of
heavy oils. J. of Can. Pet. Tech., Vol. 32, No. 9, 50-54.
Palmer I, Vaziri H, Khodaverdian M, McLennan J, Prasad K, Edwards P., Brackin C, Kutas M. 1995.
Completions and Stimulations for Coalbed Methane Wells. SPE 30012, 13p.
Poon D., Kisman K. 1991. Non-Newtonian effects on the primary production of heavy oil. Petroleum
Society of CIM/AOSTtL~ Annual Technical Meeting, Banff, Alberta.
Smith G.E. 1988. Fluid flow and sand production in heavy oil reservoirs under solution gas drive. SPE
Production Engineering, Vol. 3, No. 2, 169-180.
Squires A. 1993. Interwell tracer results and gel blocking program. 10th Annual Heavy Oil and Oil Sands
Technical Symposium.
Tremblay B., Sedgwick G., Forshner K. 1996. Simulation of cold production in heavy oil reservoirs:

ISSN 0148-9062
To cite this paper: Int. J. Rock Mech. & Min. Sci. 34:3-4, paper No. 323. Copyright © 1997 Elsevier Science Ltd

wormhole dynamics. SPE/DOE 35387, 10th Symp on Improved Oil Recovery, Tulsa, OK, 427-442.
Vaziri H., And Byrne RM. 1990. Analysis of stress, flow and stability around wellbores. Geotechnique,
40(1): 63-77.
Vaziri H. 1995. Analytical and Numerical Procedures for Analysis of Flow-Induced Cavitation in Porous
Media. Int. J. of Computers & Structures, Vol 54, No. 2, 223-238.
Vaziri H., Palmer I., Wang X. 1997a. Wellbore completion technique and geotechnical parameters
influencing gas production. Canadian Geotechnical Journal, 34:0-00 (to appear in the Feb issue)
Vaziri H., Wang X., Palmer I., Khodaverdian M., McLennan J. 1997b. Back analysis ofcoalbed strength
properties from field measurements of wellbore cavitation and methane production. In print, Int. J. of Rock
Mech. and Mining & Geomech. Abst.
Yeung K.C. 1995. Cold flow production of crude bitumen at the Burnt Lake Project, Northeastern Alberta,
Canada. 6th UNITAR Int Conf on Heavy Crude and Tar Sands, Houston, Feb 12 - 15, 3 lp

ISSN 0148-9062

Вам также может понравиться