Professional Documents
Culture Documents
SPE 206153 Ms
SPE 206153 Ms
SPE 206153 Ms
This paper was prepared for presentation at the 2021 SPE Annual Technical Conference and Exhibition held in Dubai, UAE, 21 - 23 September 2021.
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
In recent years, it has become common practice for operating companies to make a significant effort in
the planning of gravel pack installations, especially in their most important wells. Typically, the placement
of the gravel pack is accurately modelled, and multiple contingencies developed for potential alternative
scenarios to reduce the overall risk of execution. After the pack has been placed, the use of down-hole
gauge data enables the gravel pack to be fully evaluated in order to confirm success and investigate any
issues or failures. This understanding feeds into improved designs and ever higher success rates for future
completions.
The most challenging gravel packs Operators are installing today are those placed in long horizontal open
holes, around screens fitted with Inflow Control Devices (ICDs) or Autonomous Inflow Control Devices
(AICDs). Simulating gravel pack placement in wells such as these requires the effective modelling of
unusually dynamic and complex flow paths. Until recently, no simulator could adequately model these
treatments. Most jobs have also been done without the downhole gauges necessary for a complete job
evaluation, which has resulted in a lack of data for job evaluation and understanding.
Consequently, completions requiring the pack to be placed around ICD/AICD screen assemblies have,
until recently, been done without the assurance of pre-job gravel pack placement modelling. The lack of
an adequate simulator has also meant that, even on these complex and valuable wells, Operators have
been restricted to simple volumetric evaluation of their success. With no way to understand actual packing
mechanisms or investigate root causes of failures, the assumptions made on how packing occurs in these
complex wells have remained unconfirmed.
Recent evolution of gravel pack simulators has made the effective pre-job simulation, and post-job
evaluation, of gravel packs placed around ICD/AICDs a reality. This paper presents the results of the
first evaluation of a multi-proppant deep water horizontal alpha beta gravel pack around AICD screens. It
facilitates the understanding of how such gravel packs are placed, validates the packing efficiencies, and
illustrates the benefits of using multiple gravels for packing.
Introduction
The Atlanta field is an ultra-deep water (1,550 m) post salt oil field in the Santos basin offshore Brazil,
that was discovered in 2001. The Eocene reservoir has a high net-to-gross (82-94%), high permeability
2 SPE-206153-MS
(5D) and porosity (36%), and produces 14 API 228 cp oil, with a large active aquifer below it. The
sandstone is characterized as having poorly sorted non-uniform grain size. Its low burial depth (800m)
and consequent very low frac gradient (0.5 psi/ft) make the necessary horizontal open hole gravel pack
completions challenging. High-powered deep-set electric submersible pumps (ESP's) are required to assure
the commerciality of the production. The significant drilling and completion challenges presented by the
field have required technology evolution since its discovery to enable its commercial development.
Challenges
With the large aquifer below the Atlanta oil reservoir, the use of inflow control technology in the horizontal
section was seen as being beneficial for the field development. However, due to the very challenging pack
pressure placement window, the use of such technology was discarded for the first two development wells in
the field. The understanding of the minimum horizontal stress, and the attainable limits of the technologies
employed for placing HZ-OHGP's in low pressure environments, were both improved with the first two
wells; and, as a consequence of this, it was determined that a practical operational window did exist for
the application of inflow control technology for the gravel pack of the third well (Pedroso et al. 2020). As
a minimum it was expected that the screen assembly could be covered with gravel during the alpha wave
placement, providing adequate protection for the screen. AICD's were chosen for the well (Fig. 1), and it
was postulated that should the screen jacket fail, the AICD's would be plugged by any produced formation
sand thus preventing sand production.
SPE-206153-MS 3
The AICD selected was tested to confirm the associated pressure drop that would be expected under
gravel pack pumping conditions. Fig. 2 shows the pressure drop's that were determined with 1.0 cp fluid
across one insert.
The AICD design was based on expected well production, permeability variations and pressure draw-
down across the sand-face. As shown in Fig. 3, screens with 3, 6 and 8 AICD inserts were distributed along
the horizontal section.
4 SPE-206153-MS
Whilst the AICD's selectively control formation fluid inflow during production, they also present a
challenge during gravel pack placement, especially considering the very limited pack placement pressure
window of the Atlanta field. Once the optimized AICD screens configuration had been determined,
completion design strategies had to be re-considered and the pumping program re-evaluated in order to
adapt the alpha beta packing technique to this challenging scenario, and achieve an effective pack.
• Enhanced processing of the sonic calliper log was done to improve its accuracy.
• A sacrificial screen was run below the production screen assembly to allow the alpha wave to be
pumped with conventional gravel and at conventional rates.
• Since a significant pressure increase was expected during the beta wave, gradual pump rate
reductions were planned to control bottomhole pressure.
• The wash-pipe used inside the 6 5/8" screens was reduced to 4" in order to attenuate the beta wave
pressure increase seen in the modelling of pre-job simulations.
• In order to ensure screen protection and support productivity, the gravel pumping was initiated
with ceramic gravel and designed to form an alpha wave high enough to completely cover all the
production screens, before covering the sacrificial screen. Since a significant increase in pressure
was expected after covering the sacrificial screen, the volume of the ceramic gravel was calculated
to form approximately 90% of the alpha wave. The ceramic gravel was then replaced with a light
weight gravel that was more easily transported, enabling pump rates to be reduced to the operational
minimum to reduce pressure drops across the AICD screens during the beta wave. The locally
sourced light weight gravel used for this treatment was a phenolic coated ground organic material
with a nominal density of 1.34 SG.
SPE-206153-MS 5
• 3 joints of screen with 8 AICD inserts each, were run inside the casing, to facilitate beta wave pack
placement towards the heel and ensure packing into the casing. The pre-job simulations suggested
that these 24 ports would enable packing into the casing at the minimum pump rate.
• The use of friction reducer in the carrier fluid was discontinued due to its inefficiency at the low
rates expected at the end of the job, when pressure drop would be higher.
• A widely used gravel pack placement simulator was used to carefully model the alpha and beta
wave development over the irregular AICD inserts to ensure complete screen coverage and packing
Data Limitations
While several Operators have installed gravel packs around ICD/AICD screen assemblies, little has been
done to use detailed post job analysis to confirm how the actual packing proceeded. Only very limited
attempts were made to collect down-hole data on these jobs, and pressure gauges have rarely been included
in the running assemblies.
When the completion for the Atlanta well was being designed, and even installed, a simulator that
could effectively model both the packing around AICD's and the use of multiple gravels with different
characteristics was not available to the industry. As a consequence, specifications of the job data that was
collected and recorded was not determined by the requirements of a detailed post job analysis. Despite this,
downhole wash-pipe pressure gauges were run on this well and, for the first time, bottom hole data was
retrieved for the AICD gravel pack treatment.
On critical high-profile wells such as this, detailed measurement and recording of all the job parameters
would normally be recommended. This would include using multiple pressure gauges both above the service
tool and along the wash-pipe to enable a more detailed analysis of what happens during packing around
ICD's/AICD's. On this well however, just two wash-pipe pressure gauges were run: one at the toe and one
at the heel of the open hole. The analysis presented in this paper is based on the readings from surface and
these two down-hole gauges.
After pumping 2,050 bbls, a sudden drop in return rate was observed, but was not reflected in either
the surface or downhole pressure. This indicates that, in reality, there was no change in rate occurring and
the drop observed was actually an issue with the principal flowmeter. Returns were re-routed through the
back-up flow meter, and measured rates returned to expected levels. Subsequent inspection confirmed the
mechanical failure of the original flow meter.
During the job it was observed that a significant portion of the light weight gravel floated in the 9.0 ppg
(1.07 SG) brine. Investigation of this at the time of writing this paper was not possible, and so a sensitivity
study was undertaken with the simulator. It was determined that this apparent variance in the range of
particle density did not have a significant impact on the beta wave packing at the end of the job.
Modelling Challenges
The inclusion of ICD's/AICD's in an HZ-OHGP completion results in a challenging completion design, with
conflicting requirements for a low pressure drop through the screens during gravel packing, to aid placement
of the gravel pack below fracture pressure, and a higher pressure drop through the screens during production.
The flow profile created due to the presence of ICD/AICDs at every screen joint is quite complex to model,
as it requires balancing flow through the ICD/AICD, internal (screen and wash-pipe annulus) and external
(screen and open hole or casing annulus) flow paths at multiple points in the system. The longer the interval
and the more ICD/AICDs are used, the more complex the modeling becomes.
Fig. 5 shows the flow path with conventional screens (top) compared with ICD/AICD screens (bottom).
Since the fluid can easily flow through the conventional screen filter, the flowrate redistributes at each screen
joint to balance the pressure drop in the internal and external flow paths. This results in an independent
solution for each joint. However, ICD/AICD screens have a solid base pipe so the fluid can only flow
through the restricted ICD/AICD ports and the flowrate is forced to redistribute over multiple screen joints.
This results in a coupled solution for all joints with the pressure drop and flow rate across any given joint
affecting the flow profile in the other joints. The modeling solution for this requires balancing pressures
over many points (potentially hundreds) and results in a system that is very challenging to simulate.
SPE-206153-MS 7
The complexity of the flow paths in ICD/AICD gravel packs means that the ‘rules of thumb’ derived
from conventional HZ-OHGP can no longer be directly applied, and accurate modelling becomes critical
for the success of these treatments. ICD/AICD gravel pack placement designs cannot be generalized
as the performance of ICD/AICD's differs by supplier, as do their responses to changing conditions.
In order to achieve the objectives of the completion, HZ-OHGP's employing inflow control may use
different combinations of conventional screens and ICD/AICD screens, as well as ICD/AICD's with
varying configurations. A simulator that can effectively handle the complexity of ICD/AICD gravel packing
therefore becomes an important tool in their design and optimization.
Fig. 7 indicates the positions of the two wash-pipe gauges that were run on the job: one placed above the
casing shoe and the second placed close to the bottom of the screen, both reading externally. The gauges
used in the job are referred to as:
For this job, as only two externally reading wash-pipe gauges were run, only 3 section frictions are
available for analysis. Section 1 friction consists of surface line friction, work-string friction, service tool
down port friction, blank friction and AICD friction. Section 2 friction consist of screen/wash-pipe friction
in between the two wash-pipe gauges, whereas section 3 friction shows the friction in the wash-pipe, service
tool return ports and annulus above packer.
The evaluation of the treatment and the direct comparison of simulated and measured data to calibrate
the model were performed sequentially for each stage of the gravel pack treatment. The comparison of the
simulated data with the measured data are shown graphically using a solid line for measured data and dotted
line of the same color for simulated data.
10 SPE-206153-MS
Analysis Results
Circulating Step Rate Test
Calibration of the model first requires calibrating the hydrostatic at each gauge prior to the start of circulating
step rate test when the well is under static conditions. This improves the accuracy of the hydrostatic
correction and, in turn, provides a more accurate estimation of the friction pressure. The initial brine weight
had been specified as 9.20 ppg, but the observed ECD prior to the start of the circulating step rate test
As shown in Fig. 9, after calibrating all the section frictions, the measured data for each section matches
with the simulated data.
SPE-206153-MS 11
Fig. 10 shows the resulting pump pressure comparison during the circulating step rate test with the
simulated and measured data matching well. The calibrated model could then be carried forward to the
subsequent stages of the treatments.
After calibrating both reverse and circulating step rate test, the calibrated model for the full flow path
was applied to the gravel pack.
Gravel arrived below the packer after 280 bbls of pumping, but section 2 shows alpha wave deposition
starting in the open hole later at 500 bbls pumped, indicating alpha wave deposition in the blank section
from 280 bbls to 500 bbls. From 500 bbls to 2,681 bbls, the alpha wave progressed through the open hole
section with 2.71 SG, 16/20 ceramic gravel, as indicated by an increasing pressure trend in section 2. After
pumping 2,500 bbls, the gravel was switched on surface from conventional ceramic gravel to 1.34 SG, 16/20
SPE-206153-MS 13
lightweight gravel. The drop in friction at 2,681 bbls correlates to the rate reduction and, based on volumetric
analysis, arrival of 16/20 lightweight gravel at the open hole, which is consistent with the combined work-
string and blank displacement volume in Fig. 6. The alpha wave then progressed from 2,681 bbls to 3,200
bbls with light weight gravel. A significant increase in pressure is observed at 3,200 bbls which indicates
the initiation of beta wave packing, as the sacrificial screen was covered and all flow forced through the
AICDs. The rate was then gradually reduced until the end of the treatment, and section 2 friction drops
accordingly with packing continuing possibly as a combination of alpha and beta wave.
Fig. 14 shows the plot of the pack efficiency from the simulation, which matches the previous alpha and
beta wave analysis. In addition, the model indicates that the reduction in rate resulted in multiple alpha waves
between 3300 and 3500 bbls. After 3500 bbls, the beta wave re-initiates but, as the rate is continuously
dropped, a drop in pressure was observed instead of a conventional increase in pressure. The pack efficiency
also indicates an almost complete pack is achieved below 2,880 m with only some minor top side voids in the
screen open hole annulus along the horizontal extension. Although the top AICD screens above 2880m are
not fully packed, the dune height of 89% is sufficient to cover the screen and prevent any issues associated
with sand production. The overall packing efficiency obtained for this well is a very good 97%.
14 SPE-206153-MS
Fig. 15 shows a comparison of the simulated and measured wash-pipe gauge pressure data, in which
a good match is obtained. This indicates that the model accurately predicts the down-hole pressures and
packing trend during an AICD gravel pack simulation.
Fig. 16 shows a comparison of friction pressures for section 1 and section 3. The section 3 simulation
matches with the measured data throughout the gravel pack treatment. Section 1 matches with the measured
data for most of the treatment but, as soon as the rate is dropped at the start of the beta wave, a constant
pressure is observed in the measured data, whereas the simulated data shows a decrease in friction.
SPE-206153-MS 15
There are some potential explanations for this discrepancy between the simulation and the measured data.
1. Under-prediction of AICD port friction during the beta wave could be occurring. This could
possibly be due to inaccuracy of the correlations used in the model or testing limitations for which
the correlations are valid. The performance testing of the AICD's was done through a single insert,
and it is this performance that is incorporated into the model. For the job, multiple inserts were used,
and additional testing was done with 4 inserts. This test data was compared with the model and
the model closely represented the actual data, validating the implementation. However, all the tests
were done with a single AICD screen joint whereas the treatment is performed with multiple joints.
It is possible that the flow distribution through each AICD joint is not accurately modelled and this
results in the discrepancy. It is therefore recommended to test with multiple AICD screen joints in
future to determine the flow distribution and pressure drop through each AICD joint. The model
can then be compared with the actual test data to better understand the source of the differences.
2. The AICD performance is viscosity based, and viscosity has a significant impact on the pressure
drop. Brine with a viscosity of around 1 cp was used for the job. The viscosity of brine with the
addition of additives was not actually measured. It is possible that the viscosity defined in the
model did not accurately represent the actual viscosity of the fluid that was used. An increase of as
little as 1 cp in the viscosity of the brine used in the simulation, resulted in a significantly different
pressure response during the end of the beta wave.
3. Severe gravel settling in the work-string could cause the surface pressure to remain constant while
the rate was constantly dropped. Although ceramic gravel settled in the work-string during the
alpha wave, this scenario is considered unlikely, as light weight gravel was used during the beta
wave, and a large portion of the material was reported as being buoyant in the brine. Only ceramic
gravel was reversed out at the end of the job.
With the data currently available on this well, it is not possible to confirm the cause of the discrepancy. As
this is the first AICD job done with downhole gauges, the data is too limited for full validation of the model.
Additional data from future jobs will allow the model to be verified and any discrepancies investigated
in more detail such that improvements can be incorporated. Also, additional testing with multiple inserts
and screen joints must be conducted for future operations to allow the pressure drop in this scenario to be
accurately verified.
16 SPE-206153-MS
Fig. 17 shows the comparison of pump pressure between simulated data and measured data. The simulated
data matches with the measured data for most of the treatment. Only towards the end, as the rate is dropped,
the simulated pump pressure deviates from the measured data, which is due to the difference in Section 1,
as discussed above.
Acknowledgments
The authors would like to thank Enauta, Halliburton and DuneFront for permission to publish this paper.
References
C. Pedroso, et al., 2015. Packing the Un-packable : How The Use of Best Sand Control Technologies Reversed the
Marginal Project of Atlanta in a Promising Development. Presented at the SPE ATC&E, Houston, Texas, 28-30
September 2015. SPE 174892-MS
C. Pedroso, et al 2020. First Open Hole Gravel Pack with AICD in Ultra Deep Water. Presented at the SPE FDC, Lafayette,
USA, 19-21 February 2020. SPE-199325-MS
P. Wassouf, et al. 2016. Engineering Gravel Packs for Reliability. Presented at the SPE Annual Technical Conference and
Exhibition, Dubai, UAE, 26-28 September, SPE-181397-MS. https://doi.org/10.2118/181397-MS
S. Jain, et al., 2018. Optimizing Sand Control Treatments Using Gauge Data. Presented at the SPE Asia Pacific Oil & Gas
Conference, Brisbane, Australia, 23-25 October, SPE-192094-MS. https://doi.org/10.2118/192094-MS