A4.4 - Electrical Resistivity Tomography for Magnetic Flowmeter

SENSOR+TEST Conferences 2011
2011-06-07 - 2011-06-09
Proceedings SENSOR 2011
A4 - Flow Sensors I
K. Hencken, D. Schrag - ABB Switzerland Ltd, Baden Dättwil (Switzerland), H. Grothey - ABB Automation Products GmbH, Göttingen (Germany)
106 - 111


Magnetic flowmeter are successfully used for flow measurements in partially filled pipes. This is typically in waste water applications, where the tube diameter is given by the maximum flow rate, which is much larger than the usual one. In order to measure correctly the flow, the filling level of the liquid is required in addition to the induced voltage measurement. One possible filling level measurement currently used is to employ a multielectrode system and combine several voltage measurements, while a current is applied to another set of electrodes.
We investigate the possibility of using the existing multi-electrode configuration of the ABB system for electrical resistivity tomography (ERT). ERT is a method to analyze the conductivity distribution inside a closed system based on resistivity measurements at the outside and using mathematical methods to solve the inverse problem. It has found a number of applications not only in the medical area, but also for industrial and process applications. We use the open source code EIDORS in order to reconstruct the conductivity distribution in the pipe with a main focus on determining the surface of the liquid. The main limitations in our case are the small number and the asymmetric distribution of the electrodes found in the system. In order to determine the filling height from the reconstructed conductivity profile it is necessary to make use of the prior knowledge of the expected conductivity profile in the pipe. This is done by using specific “priors” in the reconstruction process, which make the ill-posed
inverse reconstruction better behaved. We have compared a number of priors, which are typically used, and have finally developed a new one, which is especially suited for the flat conductivity profile with a liquid surface in our case. We have successfully reconstructed the filling height, but find that there are major limitations given by the required computational power, as well as, for the accuracy of the filling level required for the application.
For the implementation of the tomographic measurement principle into a device with low computation capabilities we have therefore reduced the complexity of the reconstruction algorithm by using parameterized conductivity distribution profiles. A prototype system was build based on a commercial magnetic flowmeter. The system has been tested and the results show an improved accuracy compared to the traditional approach.