HR: 0830h
AN: H31B-0469 [PDF]
TI: Electrical Monitoring of Fresh Water Displacement in a Brackish Aquifer During Aquifer Storage and
Recovery: Forward and Inverse Modeling Results
AU: * Levannier, A
EM: arnaud_levannier@hotmail.com
AF: EOST, 5 rue Descartes, Strasbourg, 67000
France
AU: Delhomme, J
EM: delhomme1@slb.com
AF: Schlumberger, 1 rue Becquerel, Clamart, 92140
France
AB:
Aquifer storage and recovery (ASR) projects are now used to temporarily store water in the subsurface and to recover it when
needed. When freshwater is injected into a brackish aquifer, a transition zone forms, due to mixing, diffusion and gravity.
The front displacement and the width of the transition zone depend on the characteristics of the aquifer but, from repeated
surveys conducted with an array of downhole electrodes placed against the borehole wall, the changes in the front
position/shape can be continuously monitored.
Synthetic data were created for a targeted ASR situation through hydrodynamic and hydrodispersive modeling (performed with a
finite difference scheme) that gave the salt concentration distribution in the aquifer, as a function of space and time,
during ASR inject/store/pump cycles. Concentrations were converted first into water resistivity values {\it R$_{w}$}, and
then into formation resistivity values {\it R$_{t}$} through Archie's law {\it (1)} calibrated on logging data:
\begin{equation}
R_{t}=\frac{a}{\phi^{m}}R_w
\end{equation}
where {\it $\phi$} is the porosity, and {\it a} and {\it m} depend on the lithology.
Based on this information, the response of downhole electrodes was computed by solving equation {\it(2)} (using a finite
element modeling code) for electrical surveys conducted at repeated times during the planned ASR cycles, and in particular
during the initial ASR testing phase:
\begin{equation}
\nabla.\sigma\nabla\Phi=0
\end{equation}
where {\it $\sigma$} is the electrical conductivity and {\it $\Phi$} the potential.
The sensitivity of the electrode monitoring technique to front displacement and shape changes was first demonstrated: with a
1 Amp injected current, $\frac{\Delta V}{\Delta r}\approx1V/m$ and $\frac{\Delta V}{\Delta r}\approx50mV/m$ for the short
spacing electrodes, when the front is respectively 5 m and 35 m away from the wellbore. The inverse problem was then solved
for the synthetic electrical survey data. The misfit functional {\it(3)} was minimized under two types of constraints:
material balance (average front position being a function of the injected volume of water), and expected dispersion
(relationship between transition zone width and front position). The {\it a priori} model at each time step was taken from
the
inversion results at the previous one.
\begin{equation}
J=\|d_{p}-d\|^{2}+\alpha\|m-m_{apriori}\|^2
\end{equation}
where {\it d$_{p}$} and {\it m} are the predicted data and the model parameters, {\it d} is the synthetic (or observed) data,
and {\it m$_{apriori}$} is the {\it a priori} model taken from the previous time step.
This study showed the feasibility of using electrical monitoring for ASR projects, and notably for an early aquifer
characterization, during the initial injection phase.
UR: http://phineas.u-strasbg.fr/~arnaud
DE: 0925 Magnetic and electrical methods
DE: 1832 Groundwater transport
DE: 1894 Instruments and techniques
DE: 3210 Modeling
SC: Hydrology [H]
MN: 2003 Fall Meeting