The measurement of fluids in the oil and gas industry requires a robust measurement of multiphase flows. Magnetic resonance as a measurement principle has multiple advantages over existing technologies (one single measurement principle, measurement performed from outside the pipe with no intruding sensors, full bore design, suited for producing wells and high sensitivity at high water liquid ratios). A magnetic resonance based multiphase flow meter which is capable of producing an image of the spatial distribution of a multiphase flow has been developed. This article describes the principles of magnetic resonance. Afterwards details of the technical implementation and the method by which the system determines multiphase flow composition are explained.

No abstract

The use of reconfigurable FPGA devices to support the execution of computationally intensive software tasks is discussed in this paper. A system architecture consisting of multiple serially-connected FPGAs is developed, where each FPGA holds a pool of reconfigurable regions. An accelerator can be reconfigured into a region, replaced or discarded at runtime. Configurable connection blocks are responsible of directing data between any two accelerators. The whole system is connected via PCIe-interface to a host PC, where a middleware layer hides all hardware management operations, e.g. routing the data sent among the accelerators, and provides the end-user with an API to use the whole system. Recently, the very fast interfaces for reconfiguring parts of the used FPGAs minimize the overhead caused for hardware modifications. In addition, a manual design of hardware accelerators is not more needed with the continuously improving quality of high-level synthesis tools. In this paper, we considered the case where our system is used within MATLAB. We build a small library to compare and improve upon the execution times of some often used functions.