Abstract
Cross-borehole ground penetrating radar (XBGPR) is a geophysical technique that allows for high-resolution characterization of the interwell region. This method relies on the propagation of electromagnetic (EM) waves (typically in the MHz region) that are analyzed to generate subsurface maps of EM properties, which can be interpreted into fluid saturation maps. We present the results of a large-scale field experiment complemented by 3D numerical simulations evaluating the feasibility of locating small fluid targets of different compositions in the subsurface.
An array of 33 target-wells and 6 observation-wells completed with nonconductive, nonmagnetic pipe was used to conduct the experiments. The target-wells were filled with different fluids, including dielectrically and magnetically tagged fluids such as polymer solutions and polymer solutions with magnetite. Time-domain EM measurements were acquired using a 100 MHz cross-borehole GPR system in a semi-reciprocal tomographic setup. The acquired waveforms were filtered and processed using bh_tomo, an open source platform for cross-borehole GPR analysis. Travel-time and amplitude inversions were performed to obtain velocity and attenuation maps of the surveyed area. In parallel, 3D numerical simulations were conducted using a commercially available finite element modelling (FEM) package. The simulation results were compared and validated with the experimental results.
The simulations are in overall agreement with the field results; showing the right trend in travel-time and amplitude for the different fluids. All fluids caused an increase in travel-time compared to air-filled target-wells. Water appears to cause the largest increase, followed by AN-132, xanthan, and finally xanthan with magnetite. The observed travel-time is lower than expected. This may be an indication of the wave going around the holes and partially avoiding the slow fluids, especially because the operating wavelength is comparable to the well spacing. Another possible explanation is that the actual location of the wells is slightly different from the original design due to inaccuracy while drilling. Yet another possibility is that the array may behave as a periodic structure, causing modal propagation. The attenuation data shows a clear difference between empty and liquid-filled target holes, but little difference between the liquids. As a whole, the results prove that our approach can be used to locate relatively small fluid targets via EM tomographic surveys with no previous geological information.
Experimental data of cross-borehole GPR is rather limited. Our experiments expand the understanding of the challenges and opportunities that such technique can offer to the oil and gas industry. We have also developed and validated modelling capabilities that will enable improved planning and quick testing of future surveys.
