Summary
A significant proportion of future oil production is expected to be driven
by water injectors in reservoirs that are sand prone. Achieving sweep
efficiency and sand control in such formations is challenging. In many cases,
the ideal sand control is no sand control [e.g., a cased-and-perforated
(C&P) completion] that requires rigorous sanding assessment. Sand
production in injectors often goes unnoticed until it is too late (sand
covering the pay), making it difficult to ascertain the specific set of
conditions resulting in sanding and the severity of the individual sanding
episodes. On the basis of physics and mechanisms governing sanding, general
nonquantitative factors can be postulated on the causes of sanding. To provide
a deeper insight into this matter, a numerical study has been undertaken to
model sanding in injectors, accounting for several intercoupled factors,
including, among others, injection pressure, crossflow, water hammer (WH)
pressure pulses, and degradation of the formation matrix resulting from
repeated shutdowns.
This paper describes the concepts used for sand-production modeling and
shows application of the model to a field problem involving a C&P
completion in a sand-prone reservoir. The results show that the mode and
magnitude of sanding are influenced by the rock properties, injection
operations, and the equipment type and installation. The cases analyzed
indicate a correspondence between the rate of shut-in and the onset of sanding.
In cases involving unconsolidated sands, the WH effects have a pronounced
impact on sanding. Sand control can be omitted in even extremely weak rocks if
the injection pressure is optimized, frequency of hard shutdowns is controlled,
and hardware is positioned in a manner that reduces the WH-pressure-pulse
magnitude. The proposed modeling can be used when determining the sand-sump
capacity required over the projected life of the well.
Introduction
In comparison with the vast reported studies devoted to sand production in
producers since the early 1980s (Bianco and Halleck 2001; Nouri et al. 2006;
Papamichos and Malmanger 1999; Risnes et al. 1982; Tronvoll et al. 1992; Veeken
et al. 1991; Vaziri et al. 2006), little attention has been given to sanding in
injector wells. There are considerable differences in sanding between injectors
and producers. Generally, high depletion and drawdown may lead to rock
disaggregation in the latter, which is a necessary condition for sanding.
Thereafter, the bean-up strategy, drawdown magnitude, and water cut are the
predominant factors influencing sanding (Vaziri et al. 2006).
In water injectors, there is no depletion, so any rock disaggregation would
be the result of near-wellbore degradation brought about by high injection,
which may lead to fracturing and/or shear failure and the pressure cycles
associated with repeated shutdowns and WH effects. Potential for sanding
episodes occurs over only the short periods immediately after a shutdown while
crossflow and/or backflow, along with any WH pulses, remain active.
Whereas in producers any disaggregated sand near the wellbore can become
compacted with drawdown, in injectors, the near-wellbore sand is likely to be
under a very low effective-stress state and thus, be quite susceptible to
production. Therefore, the pressure gradient required to create sanding in
injectors upon shutdown is expected to be much smaller than that under normal
production operation.
Unless flow rates are too insufficient to transport sand to the surface in
producers, any sand production can be detected usually by a variety of methods,
such as collecting samples at the surface and sand detectors, and other
evidence (sand in separators, erosion of tubing along the production line and
in other hardware such as chokes). This enables the operators to change
strategies to mitigate sanding and monitor future activities more closely. In
injectors, on the other hand, any sand will go down to the bottom of the well,
and the first real evidence of its occurrence typically surfaces with a
significant loss in the injection performance, which is a strong indication of
sand covering the pay interval. Owing to this, little direct field data are
available on when sanding in injectors starts, how much sand is brought in with
each shutdown, and which factors play the key role in assisting the development
of effective sand-mitigation practices. Among the rather sparse literature on
sanding in injectors are Morita et al. (1998) and Santarelli et al. (2000).
On the basis of field observations, Morita et al. (1998) linked sand
production in injector wells to repeated shutdowns. They reviewed the
injectivity efficiency of several completion options and provided guidelines
for selecting completion methods for water-injector wells drilled in weak
formations. Santarelli et al. (2000) examined a number of water injectors in
the North Sea that had experienced extreme losses of injectivity over
relatively short periods of time. They discussed several potential factors
influencing sanding and well-injectivity loss but without quantifying these
factors with measurements or rigorous analyses.
The objective of this paper is to quantify the role of various factors that
are known to influence, or suspected of influencing, sanding in water
injectors. These include the injection pressure, frequency and rate of
shutdowns, magnitude of inflow caused by backflow or crossflow, and WH pressure
pulses. A customized finite-difference numerical model is used for this study.
This model uses fully coupled fluid-flow and formation-stress formulations and
allows for material degradation caused by stress/strain cycles and seepage
forces, including fluidization; it also has built-in criteria for simulating
conditions resulting in sanding and tracking the behavior of the formation
following sand production. Hence, computation of sanding can be performed over
the life of the well and not simply in detecting the onset of sanding. This
issue is of particular practical interest because, unlike producers, sand
production in injectors can be tolerated, provided it does not compromise
injectivity performance (i.e., it remains within the capacity of the sump). In
this regard, modeling can show the impact of measures that can be taken to
minimize sanding (e.g., by means of reduction in the WH pressure), and it can
compute the overall sanding for a range of viable scenarios that can then be
used to assess whether an adequate sump can be created. If not, considerations
can be directed objectively toward selection of sand controls. Use of sand
control in injectors does not quite remove all concerns aside from the higher
capital expenditures (CAPEX) . Compared to C&P completions, they are quite
susceptible to plugging from any impurities in the quality of the injected
water, and from formation fines that can be quite mobile immediately after a
hard shutdown. Loss of injectivity is a major concern.
This paper provides a quantitative insight into the role of several factors
that impact sanding in injectors, which can be used to develop effective
practices and measures to reduce sanding severity. While the problem
definition, properties selected, and the boundary conditions for the cases
shown are based on real field conditions, the sanding predictions have not been
validated because of the current lack of ability to monitor sanding in the
field.
© 2008. Society of Petroleum Engineers
View full textPDF
(
1,388 KB
)
History
- Original manuscript received:
27 February 2007
- Meeting paper published:
30 May 2007
- Revised manuscript received:
7 April 2008
- Manuscript approved:
7 April 2008
- Version of record:
15 November 2008
View Cart
