Hydrogeological application of electrical resistivity tomography, implementing a fixed-electrode strategy

Abstract

The requirements of environmental assessments and of understanding and monitoring in-situ mass and heat processes in porous media have led to the development of geophysical methods for remote mapping and monitoring of contaminant plumes and fluid migration. With the possible exception of seismic approaches, electrical methods known as Electrical Resistivity Tomography (ERT) have become the most widely studied and used for these purposes. Wherever a sufficient contrast in ground resistivity is generated by human or natural processes, monitoring the resistivity structure over time may give insight into these processes. ERT has monitoring applications in processes such as Enhanced Oil Recovery (EOR), Slurry Fracture Injection 9SFI), and monitoring transport processes in hydrogeology. A permanent electrode arrangement for long term monitoring removes the effects of Earth's heterogeneity and anisotropy when a process is analyzed as a function of time.

As a starting point on the work described in this thesis, ERT data were collected from a Cambridge, Ontario, sand pit before, immediately after and one week following a 11000 liters slurry injection. These measurements verified that ERT could detect changes caused by the injection and later movement of this conductive mixture in the ground. The commercial equipment used for these measurements was not well suited to the tasks, mainly because it was extremely slow. Further, there was a lack of robust and user-friendly three-dimensional modeling software to use as a means of predicting response and -eventually- as the engine of an inversion routine. Finally, it was difficult to analyze the injection situation in terms of how best to place a limited number of surface and borehole electrodes to most effectively monitor the injection fluids. The remainder of the thesis addresses these problems.

The first objective was to design and construct a more suitable ERT measurement system.

The image resolution of the basic ERT technique is usually poor. Given the normal limitations of excitation current and geology, there are three approaches to improving the resolution of these images: - Increasing the accuracy and precision of the measuring instruments - Using optimal electrode arrangement with respect to the resistivity anomaly in question - Processing data with methods that eliminate noise and improve the resolution of the resistivity (conductivity) distribution map.

With an appropriate measuring system, it should be possible to increase the speed, repeatability, and accuracy of ERT data collection considerably. The system described here consists of the following: i. Expandable network switches for signal input and output; ii. Low-pass filter (10 Hz) for noise reduction; iii. Programmable gain for efficient data collection; iii. A 16-channel, 16-bit A/D board having software for range control; and, iv. A proper field computer system (i.e., a minimum of 80486 PC).

Since the stabilized portion of each of positive and negative induced current pulses in each cycle is about 0.4 second in a 0.5 Hz current, it is possible to collect the voltage differences of an electrode array in a mattery of minutes.

The second objective was to adapt SALTFLOW as a platform for both the resistivity and hydrogeological modeling of the saline groundwater flow resulting from waste injection. Because steady-state groundwater and direct electrical current flow obey the same governing equations, it has been possible to make use of the modeling expertise and one of the software programs available through the Waterloo Groundwater Simulation Group (SALTFLOW and FRAC3DVS) to experiment with three-dimensional ERT monitoring scenarios.

Further improvement of the forward model have been achieved by compared a combination of different iterative sparse solver methods (such as: Conjugate Gradient (CG), Bi-Conjugate Gradient (BICG), Conjugate Gradient Stabilized (CGStab), Generalized Minimum Residual (GMRES)) and four ordering methods (natural, RCM, Min Degree, Nested dissection). Considering the same resolution and tolerance, it has been determined that the combination of nested dissection ordering with preconditioned conjugate gradient method shows faster convergence. Implementing line elements for current electrode in the forward model, which has been used for groundwater modeling (as a highly conductive line element to represent wheels), reduces the oscillation of the model results on the grids around the current electrodes.

The third objective was to develop methods of sensitivity analysis that will allow a more efficient examination of the electrode arrays that could be effectively used in a given situation.

The sensitivity analysis is based on the state sensitivity and adjoint sensitivity techniques. These have been implemented in the forward model.

The electrode setup is one of the most important issues in an ERT survey. The closer the electrodes are to the anomaly, the better will be the quality of the ERT images; however, the optimal electrode arrangement depends not just on the target, but also on the field conditions, the availability of boreholes, project budget and the client's expectations. We are using sensitivity analysis and an improved forward model to estimate the optimal electrode placement for a given target under given circumstances.

The fourth objective was to demonstrate the ERT method and the improvements undertaken by the author on the data collected at the Cambridge injection site.

The thesis has not, in fact, met all these objectives, but has made substantial progress towards them. The complete design of the measurement system and the construction of its potential measurement components were achieved. A lack of capacity in the science shops, however, resulted in the power (current) supply not being constructed in time for field evaluation of the injection or its aftermath.