Calibration of electrical impedance tomography

Daily, W ; Ramirez, A

Lawrence Livermore National Laboratory

关键词: Calibration;    Tomography;    Mineralization;    Resistors;    Electrodes;   

DOI  :  10.2172/15005726
RP-ID  :  UCRL-ID-139413
RP-ID  :  W-7405-ENG-48
RP-ID  :  15005726

美国|英语

来源: UNT Digital Library

【 摘 要 】

Over the past 10 years we have developed methods for imaging the electrical resistivity of soil and rock formations. These technologies have been called electrical resistance tomography of ERT (e.g. Daily and Owen, 1991). Recently we have been striving to extend this capability to include images of electric impedance--with a new nomenclature of electrical impedance tomography or EIT (Ramirez et al., 1999). Electrical impedance is simply a generalization of resistance. Whereas resistance is the zero frequency ratio of voltage and current, impedance includes both the magnitude and phase relationship between voltage and current at frequency. This phase and its frequency behavior is closely related to what in geophysics is called induced polarization or (Sumner, 1976). Why is this phase or IP important? IP is known to be related to many physical phenomena of importance so that image of IP will be maps of such things as mineralization and cation exchange IP (Marshall and Madden, 1959). Also, it is likely that IP, used in conjunction with resistivity, will yield information about the subsurface that can not be obtained by either piece of information separately. In order to define the accuracy of our technologies to image impedance we have constructed a physical model of known impedance that can be used as a calibration standard. It consists of 616 resistors, along with some capacitors to provide the reactive response, arranged in a three dimensional structure as in figure 1. Figure 2 shows the construction of the network and defines the coordinate system used to describe it. This network of components is a bounded and discrete version of the unbounded and continuous medium with which we normally work (the subsurface). The network has several desirable qualities: (1) The impedance values are known (to the accuracy of the component values). (2) The component values and their 3D distribution is easily controlled. (3) Error associated with electrode noise is eliminated. (4) Each box formed by 12 adjacent components corresponds to a voxel in the finite difference forward model used in the inverse code and this correspondence makes for easy comparison of inversion results and model physical parameters. Using this network we can study the errors associated with the measurement system (called Zombie) separated from the errors introduced by electrode noise. We can also learn details in the behavior of the inversion software (called CR3D) by comparing images and model.