A dual-mode tomography system based on electrical capacitance and gamma-ray tomography has been developed at the Department of Physics and Technology, University of Bergen. The objective of the dual-mode tomograph is to acquire cross-sectional images, i.e. tomograms, of hydrocarbon flow comprising oil, water and gas constituents. The capacitance tomograph utilizes an eight-electrode sensor set-up mounted around a PVC pipe structure which is sensitive to the electrical permittivity εr of the fluid. By using the capacitance tomograph, the produced water constituent can be separated from the gas and crude oil constituents, assuming that the liquid phase is oil continuous. The high-speed gamma-ray tomograph comprises five 500 mCi 241Am gamma-ray sources, each at a principal energy of 59.5 keV, which corresponds to five detector modules, each consisting of 17 CdZnTe detectors mounted around the same pipe section as the capacitance sensor. The gamma-ray tomograph discriminates between the gas and the liquid phase, since these have different photon attenuation properties. As a result, the gamma-ray tomograph is able to distinguish the gas phase from the liquid phase of the hydrocarbon flow. Consequently, the dual-mode capacitance and gamma-ray tomography set-up is able to distinguish the oil, water and gas constituents of hydrocarbon flow. This paper presents the work that has been performed related to static characterization of the dual-mode tomograph using the Landweber reconstruction algorithm on polypropylene phantoms. The objective of the work has been to quantitatively evaluate the static imaging performance of the dual-mode tomograph with respect to relative spatial measurement errors, i.e. root mean square errors of the reconstructed tomograms compared to that of the phantom. The work shows that dual-mode tomography using electrical capacitance and gamma-ray sensors is feasible on hydrocarbon flow components using a pixel-to-pixel fusion procedure on separately reconstructed tomograms based on the Landweber reconstruction algorithm.

A data acquisition and control system (DACS) for high-speed gamma-ray tomography based on the USB (Universal Serial Bus) and Ethernet communication protocols has been designed and implemented. The high-speed gamma-ray tomograph comprises five 500 mCi 241 Am gamma-ray sources, each at a principal energy of 59.5 keV, which corresponds to five detector modules, each consisting of 17 CdZnTe detectors. The DACS design is based on Microchip's PIC18F4550 and PIC18F4620 microcontrollers, which facilitates an USB 2.0 interface protocol and an Ethernet (IEEE 802.3) interface protocol, respectively. By implementing the USB-and Ethernet-based DACS, a sufficiently high data acquisition rate is obtained and no dedicated hardware installation is required for the data acquisition computer, assuming that it is already equipped with a standard USB and/or Ethernet port. The API (Application Programming Interface) for the DACS is founded on the National Instrument's LabVIEW R graphical development tool, which provides a simple and robust foundation for further application software developments for the tomograph. The data acquisition interval, i.e. the integration time, of the high-speed gamma-ray tomograph is user selectable and is a function of the statistical measurement accuracy required for the specific application. The bandwidth of the DACS is 85 kBytes s −1 for the USB communication protocol and 28 kBytes s −1 for the Ethernet protocol. When using the iterative least square technique reconstruction algorithm with a 1 ms integration time, the USB-based DACS provides an online image update rate of 38 Hz, i.e. 38 frames per second, whereas 31 Hz for the Ethernet-based DACS. The off-line image update rate (storage to disk) for the USB-based DACS is 278 Hz using a 1 ms integration time. Initial characterization of the high-speed gamma-ray tomograph using the DACS on polypropylene phantoms is presented in the paper.

High-speed gamma-ray tomography (HSGT) based on multiple fan-beam collimated radioisotope sources has proved to be an efficient and fast method for cross sectional imaging of the dynamics in different industrial processes. The objective of the tomography system described here is to identify the flow regime of gas/liquid pipe flows. The performance of such systems is characterized by the spatial resolution, the speed of response and the measurement resolution of the attenuation coefficient. The work presented here is an experimental analysis of how the measurement geometry and the reconstruction method affect the error of the reconstructed pixel values. These relationships are well established for medical x-ray tomography where high intensity x-ray tubes are used as sources. For radioisotope sources, however, the radiation intensity is limited, which causes the measurement uncertainty, i.e. the Poisson noise, to be considerably higher. In addition, the influence of scattered radiation is more severe in a multiple source radioisotope system compared to that of x-ray systems. A computer-controlled flexible geometry gamma-ray tomograph has been developed to acquire experimental data for different fan-beam measurement geometries, and these data have subsequently been used for image reconstruction using seven different iterative image reconstruction algorithms. The results show that the reconstruction algorithms perform cross sectional images with different quality and that there is virtually nothing to be gained by using more than seven sources for flow regime classification of multiphase pipe flow consisting of gas, oil and water.

No abstract