SPE 84262

Sand Control Completion Reliability and Failure Rate Comparison With a

Multi-Thousand Well Database
George E. King, Pat J. Wildt, Eamonn O’Connell, BP America Inc.

Copyright 2003, Society of Petroleum Engineers Inc.

comparison between sand control types, based on well
This paper was prepared for presentation at the SPE Annual Technical Conference and location and age of completion was not available.
Exhibition held in Denver, Colorado, U.S.A., 5 – 8 October 2003.

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, as
After consultation with various asset teams within BP
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to and even in other companies, we decided to use inter-
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at and intra-company networking to gather general sand
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
control completion performance information that could
for commercial purposes without the written consent of the Society of Petroleum Engineers is be used to do high level comparisons of sand control
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 reliability. The sand control completion reliability
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.
database was initiated with Gulf of Mexico (GoM) data
from BP shelf operations and expanded with data from
Angola, Indonesia, North Sea, Mid East and other areas.
Abstract Deepwater completions data was added as it
Through inter- and intra-company cooperation, an Excel became available.
database has been constructed with sand control
reliability/failure information on over 2000 wells. The This sand control reliability database has always been
database makes possible the study of completion free and open to participation from any company or
performance and the comparison of reliability for various group. Its general operation is governed by some very
forms of sand control, and common causes for sand basic guidelines:
control completion failures. This paper shares the 1. The well data may come from any company as
general reliability results of this database without any long as all the data for all the wells in a field
reference to producing company name, service provider are reported.
names or product names. The database contains 2. Recent data with modern completion techniques
information regarding general reliability and has were preferred over historical information with
information from many wells on cumulative production, early completion systems. A completion cut-off
maximum rates, drawdown, skins, pay deviation, screen of the late 1980’s was generally applied.
length, and pay interval sorting, fines content, depth and 3. Any company that contributed data would be
length. The database is open to any company willing to entitled to the full database, free of charge, with
share anonymous well data. There are no costs, no field, well and company names removed.
contracts and no meetings. 4. There are no agreements, contracts, lawyers,
fees or meetings.
Introduction 5. All well data that can be freely collected is
During planning for BP’s deepwater fields and many reported, but every attempt is made to omit
other developments where sand control completions are product brand names, service provider names,
considered necessary, there have been considerable and other identifiers. Comparison of products or
discussions on the type of sand control completion to service companies was not in the scope of
use for specific areas and well types. Since many of the the work.
wells were sub-sea wellhead operations, the 6. Erroneous data points and/or well information is
appropriateness of the sand control completion decision expected. Any accuracy of the database is
is made more critical by the economic necessity to attempted through sheer number of wells in
minimize, or preferably eliminate well re-entry to repair a population.
or re-stimulate these wells after the initial completion 7. No data interpretation, warranty, guarantee or
operation. There were many competing forms of sand accuracy on the data is given or implied.
control reliability information, both from service providers
and operators, but a direct, relatively unbiased Although technically simplistic in construction and intent,
the sand control reliability database has performed well
2 SPE 84262
for BP, delivering the high-level reliability estimates it
was intended to provide.
Discussion
The type, accuracy and extent of data required for a
reliability study depends on the level of detail and the
accuracy needed. Within the confines of well data
measurements over the last ten years, general
technique comparisons are possible, but consideration
of the data source is required before reliability
observations can be discussed. There are literally so
many influences on completion quality that direct well-to-
well comparisons are difficult.
Given the limitations of well comparisons, having a large
population of wells improves the probability that broad
observations from the database are more accurate.
These are a few observations from the database.
1. Failure data and failure perception are both
influenced by the data gathering process; i.e.,
some data is biased by a person’s views and the
way the data is collected. One protection from
gross error is to have a very large population of
data and to search the data for specific
information that is relative to a set of conditions.
Trust may be placed in the large data file, but
specific data points must be suspect.
2. The definition of failure is frequently disputed.
Several wells are considered successes by their
operators, but are choked back to prevent the
flow of gravel or formation sand. The sand
control in these wells has failed.
3. Short lifetime production zones skew the time
based reliability data towards longer perceived
reliability – the completion may not be
challenged by time in place, large cumulative
produced volumes or high production
rate through-put.
4. Large numbers of new wells in a new field skew
the production reliability data towards longer
perceived reliability or skew it towards shorter
rleaibility life if there are several early failures.
The most accurate picture of a completion’s
reliability comes from a large population of wells
that have produced for a long period of time.
5. Limited well numbers (fewer than 100 wells in a
population) may skew the data in any direction.
6. Failure frequency is NOT intervention frequency.
Failed wells are often allowed to “stumble along”
for years at reduced rates to avoid workovers.
7. Phase change or other strategic event during
production may cause a sudden failure.
The data was collected in an Excel spreadsheet
template that asks for type of completion, completion
and abandonment dates, well physical measurements,
deviation, formation data, various production and
performance data, skins, etc., and finally the type of
failure if one had occurred. Efforts are made to collect as
much data as possible, but all of the data records are
missing some data.
Sand control completion failures were defined (Table 1)
in an attempt to improve comparisons between
completion types and special conditions presented by
the well or the formation. There are disagreements over
what constitutes failure, but having a constant definition
assists the comparison of well performance and
reliability longevity. The failure types are design failure,
applications failure, infant failure and production failure.
The actual data records incorporate over a million
potential data cells (a fraction of which are filled) and are
too voluminous to include here. A synopsis of the data
with general failure descriptions shown in either failures
per well per year or as a percent of completions that fail
are shown in Table 2. Note that this data is not sorted for
specific conditions.
A general examination of the failures has proved
interesting in that several common failure causes could
be inferred. Some failures are independent and some
are in concert with other actions or failure causes. Note
that some root failure causes are common to all failure
types and that there is considerable influence exerted on
infant and production failures by errors in the design and
application phases.
Some Design Failure Causes:
1. Limited contact (e.g., wellbore deviation relative
to frac plane, or completion type limits formation
input – too little flow area.)
2. Too thin (<3/4”) a clearance between
screen/casing or screen/hole.
3. Choke in pathway (e.g., gravel too small for
sand, screen too small or too easily plugged)
4. Ignoring high permeability streaks - early gravel
bridging halts job
5. Impingement erosion (insufficient gravel
reserve planned)
Some Application Failure Causes:
1. Quality control problems with equipment, fluids,
tools and people.
2. Not getting perforations clean before packing or
frac packing.
3. Very low volumes of gravel displaced (probable
voids in the pack).
4. Early screenouts (bridge development) across
high permeability streaks.
5. Mixing gravel and formation sand – tool
swabbing, poor cleanup, etc.
6. Limited zone contact – perf guns
firing incomplete
SPE 84262 3

7. Screen running damage – physical damage and
particle plugging.
8. Screen collapse or abrasion during high
rate pumping.
9. Choke in the pathway from mixing gravel and
formation sand.
10. Poor gravel placement (e.g., lack of gravel, gel,
x-linker or interruption in pumping)
Some Infant failure causes:
1. Early screen plugging caused by poor
wellbore cleanup.
2. Incomplete packing or voids in the pack.
3. Bring well on too quickly – indirectly supported
by data.
General Observations:
From a general observation standpoint, database failure
causes supports the premise that formation character
and well needs must drive selection of the sand
control type:
1. Layered pay flow behavior places significantly
different “loads” on a sand control completion
than does a clean blocky sand.
2. Vertical well flow behavior and reservoir
connection may be much different than
deviated wells.
3. High rate wells place different loads on a
completion and appear to have somewhat
higher failure rates than low rate wells.
4. Sand control completion failures appear related
to local (measured) flux bpd/ft through
the screen.
5. Average drawdown has no apparent relationship
to early failure in the general data set, however
some data suggest that rapid start-ups do
influence failure in some cases.
6. If formation fines are mobile, completions will
either increase in skin over time or pass a
significant amount of fines, depending on gravel
and screen sizing.
7. Phase change during production is still a major
precursor to sand control failure, although the
largest impact of water “breakthrough” in
causing failures appears to be in cased hole
gravel packs.
8. Correct design basics and application (gravel
reserve or amount of gravel displaced,
screen/casing or screen/hole clearance, screen
aperture, hole shape, etc.) are very important.
lessening erosion by fluids and by mobile fines. The
effectiveness of the frac pack as an optimum completion
may be less effective as formation permeability
increased and the formation matrix flow contribution
became more pronounced. In high kh formations and
especially in highly deviated wells with large reservoir
contacts in blocky sands, the open hole gravel packs
appear to reliably deliver high rates with relatively low
failures. One key to OHGP performance may be the
very large wellbore flow area of the OHGP – theoretical
100% of wellbore compared to theoretical 4 to 8% open
entrance hole area in a cased and
perforated completion.
Plotting the age related failures has demonstrated that
the failures assume a rough “bath tub” failure mode with
most failures coming either very early (infant failures) or
at the “old age” failure point characteristic of the
completion type and the formation conditions.
Conclusions
The data within the database, can be sorted by general
geographic area and appears to be sufficient for high
level comparisons of sand control completion reliability.
The database is available free of charge to any company
or organization that contributes data.
Acknowledgements
The authors thank BP America Inc., for permission to
publish. We thank numerous BP BU engineers, and
members of the EPT Completions Team for their data
collection assistance. We also recognize 40 world-wide,
major and independent companies, that have directly or
indirectly donated data from operated wells, partnered
wells or public and published field-study literature.

In broad terms, frac packs appear to work better on

layered or laminated, lower permeability (<100 to 200
md) formations with reasonably vertical wellbore
sections across the pay zone. Part of this high reliability
was some companies’ expertise in designing and
applying the job. Another large part of the success was
that the flow in a fracture may, at least slightly, spread
out the high flux inflow from a high permeability streak,
decreasing the local flux rate on the screen and
4 SPE 84262

Table 1: Failure Definitions Suggested to Data Preparers.

Completion Type Design Failure Applications Infant Failure Production Failure
Failure
Within 30 days of After 30 days of start
start of production, of production,
continuously flows continuously flows
formation sand formation sand
Cased and (>screen aperture) to (>screen aperture) to
Perforated surface surface
Error in screen
diameter, length, or
slot/weave size Screen running Same as Cased and Same as Cased and
Screen Only (SOC) opening damage Perforated Perforated
Same as SOC, plus
less than 100%
screen/casing
Same as SOC, or annular gravel
error in gravel volume displaced;
volume calculation any gravel bridge in Same as Cased and Same as Cased and
CHGP or measurement, equipment Perforated Perforated
Same as SOC, plus
less than 100%
Same as SOC, or OH/screen annular
error in gravel volume displaced;
volume calculation any gravel bridge in Same as Cased and Same as Cased and
OHGP or measurement, equipment; Perforated Perforated
Same as SOC, plus
less than 100%
Same as SOC, or screen/casing
error in gravel annular volume
volume calculation displaced, any gravel Same as Cased and Same as Cased and
HRWP or measurement, bridge in equipment Perforated Perforated
Same a SOC, plus
error in prop or
gravel volume calc Same as HRWP, also
or measurement. applications related
Error in fluid leakoff failure to displace
prediction, pad vol, designed proppant Same as Cased and Same as Cased and
Frac Pack etc. load Perforated Perforated
SPE 84262 5

Table 2: General Reliability Summary on All Sand Control Completion Types

Sand
Total Sand Sand Control
number Control Sand Control Control Production
wells of Design Applications Infant Failure
Type of completion Total Well- Failure % of failure % of Failure % (failures /
Completion type Years attempts attempts of attempts well/yr
Injectors
(soft sand
formations) 24 62 0 12.5 0 0.064
Screenless
Fracs 7 44 0 0 0 0.046
Cased and
Perfed 66 306 0 1.5 0 0.085
Screen Only
Completion 183 783 0.6 0.0 0.6 0.056
Expandable
Screens 194 255 1 3 1 0.016
Cased Hole
Gravel Pack 369 1514 0 2.2 0.8 0.011
Open Hole
Gravel Pack 175 507 0 9.7 0.57 0.020
High Rate
Water Packs 187 544 0 0.5 0.53 0.009
Frac Pack 844 3369 1.69 2.4 0.24 0.004

Total 2049 wells 7384 yrs