This paper presents a new procedure to determine interwell connectivity in a
reservoir on the basis of fluctuations of bottomhole pressure of both injectors
and producers in a waterflood. The method uses a constrained multivariate
linear-regression (MLR) analysis to obtain information about permeability
trends, channels, and barriers.
Previous authors applied the same analysis to injection and production rates
to infer connectivity between wells. In order to obtain good results, however,
they applied various diffusivity filters to the flow-rate data to account for
the time lags and the attenuation. This was a tedious process that requires
subjective judgment. Shut-in periods in the data, usually unavoidable when a
large number of data points were used, created significant errors in the
results and were often eliminated from the analysis.
This new method yielded better results compared with the results obtained
when production data were used. Its advantages include: (1) no diffusivity
filters needed for the analysis, (2) minimal number of data points required to
obtain good results, (3) and flexible plan to collect data because all
constraints can be controlled at the surface. The new procedure was tested by
use of a numerical reservoir simulator. Thus, different cases were run on two
fields, one with five injectors and four producers and the other with 25
injectors and 16 producers.
For a large waterflood system, multiple wells are present and most of them
are active at the same time. In this case, pulse tests or interference tests
between two wells are difficult to conduct because the signal can be distorted
by other active wells in the reservoir. In the proposed method, interwell
connectivity can be obtained quantitatively from multiwell pressure
fluctuations without running interference tests.
Well testing is a common and important tool of reservoir characterization.
Many well-testing methods have been developed in order to obtain various
reservoir properties. Interference tests and pulse tests are used to quantify
communication between wells. These methods are often applied to two wells such
that one well sending the signals (by changing flow rates) and the other is
receiving them (Lee et al. 2003). For a large field such as a waterflood
system, however, multiple wells are present, and most of them are active at the
same time. In that case, pulse tests or interference tests between two wells
are difficult to conduct because the signal can be distorted by other active
wells in the reservoir. In this method, data can be obtained from multiwell
pressure tests that resemble interference tests. Thus, we can have several
wells sending signals and the others receiving the signals at the same time.
The wells that are receiving the signal, however, can either be shut in or kept
at constant producing rates.
The pressures at all wells are recorded simultaneously within a constant time
interval. The length of the test will depend on the length of the time interval
and the number of data points. Results of this method can be used to optimize
operations and economics and enhance oil recovery of existing waterfloods by
changing well patterns, changing injection rates, recompletion of wells, and
This work is based on previous work conducted by Albertoni and Lake (2003)
by use of injection and production rates. In their work, Albertoni and Lake
developed and tested different approaches by use of constrained MLR analysis
with a numerical simulator and then applied it to a waterflooded field in
Argentina. They used diffusivity filters to account for the time lag and
attenuation of the data. In his thesis, Dinh (2003) verified the method by use
of a different reservoir simulator and applied it to a waterflooded field in
Nowata, Oklahoma. He also investigated the effect of shut-in periods and
vertical distances on the results.
The main objectives of this work are to verify the results obtained from
pressure data with results from flow-rate data to propose a new method to
determine interwell connectivity and to suggest further research and study on
Similar to the method that uses production rates, we will concentrate on a
waterflood system only. The reservoir is considered as a system that processes
a stimulus (i.e., a well that is sending signals) and returns a response (i.e.,
a well that is receiving the signals). The effect of the reservoir on the input
signal will depend on the location and the orientation of each
stimulus/response pair. Because the total pressure changes at active and
observation wells are not equal, only the MLR (Albertoni and Lake 2003; Dinh
2003; Albertoni 2002) was used. The effect of diffusion was not significant,
thus the diffusivity filters were not used.
The method was applied to two synthetic fields, one with five injectors and
four producers and the other with 25 injectors and 16 producers.
© 2008. Society of Petroleum Engineers
View full textPDF
- Original manuscript received:
23 January 2007
- Meeting paper published:
31 March 2007
- Revised manuscript received:
20 February 2008
- Manuscript approved:
11 March 2008
- Version of record:
25 October 2008