Density Measurement of Multiphase Pipe Flows

Density is an important physical property and its measurement has wide application in a vast number of industries including; oil and gas, petrochemical, pharmaceutical, brewing, food & beverage production and mining. Density is often required to be accurately measured either as a standalone property or in combination with other flow properties for the purpose of quality assessment, process control, and custody transfer. Given the increasing importance of density measurement, extensive research has been conducted over recent years to develop newer and more accurate density measurement sensors and to improve the accuracy of existing sensors.

This thesis describes the design and development of a novel, non-invasive, non-radioactive Vibrating Density Measurement System (VDMS) capable of measuring fluid density in both single phase and multiphase flows. The device is also capable of measuring mean in-situ phase volume fractions in two-phase flows. The VDMS comprises three sub-units; (i) a measurement unit which includes a straight length of sensing pipe with corrugated bellows at both ends, an actuator and relevant sensors; (ii) a signal conditioning and processing unit; and (iii) a data acquisition unit.

The thesis also reports the development of a novel mathematical density prediction model which is used in conjunction with the VDMS. It then goes on to report the results of static bench test experiments that were conducted on the VDMS using a bespoke test rig. These tests were performed (i) to obtain the mass of the sensing pipe, its stiffness constant and damping constant; (ii) to investigate the frequency response characteristics of the VDMS; (iii) to obtain the VDMS constant; (iv) to investigate the sensing pipe displacement pattern; (v) to investigate the capability of the VDMS to give accurate density measurements of static fluids and (vi) to define the optimal VDMS operating conditions. Computation of density was achieved using the VDMS, the density prediction model and a novel signal processing technique. This signal processing technique used the Discrete Fourier Transforms (DFTs) of the measured force, used to mechanically excite the sensing pipe at its centre, and the measured displacement at the sensing pipe centre.

Next, the thesis reports a novel computer based control system that was developed to ensure that the VDMS automatically operated at its optimal operating conditions, so that errors in the density measurement were minimised. The control system was also capable of providing online computation of the flow mixture density.

Results are reported of several experiments conducted by the author on the VDMS to measure the fluid density in a range of “water only” flows, “solids-in-water” flows and “air-in-water” flows. These flows were all vertically upward and were established in the working section of a multiphase flow loop. These multiphase flow experiments were subsequently extended to compute the mean in-situ phase volume fraction of the solids phase in “solids-water” flows and the gas phase in “air-water” flows. For the “water only” and “solids-water” flow experiments, the mean error in the predicted density was consistently within 0.5% of the reference density and the standard deviation of the error was less than 1%. For “air-in-water” flows the predicted density was within 1% of the reference density for flows where the air volume fraction in the mixture was less than 10%.

The mean in-situ volume fractions, measured by the VDMS, of the dispersed solids in “solids-in-water” flows and of the dispersed air in “air-in-water” flows were within 10% of the reference measurements for the vast majority of measurements taken.

Finally, the thesis describes a modified mathematical model, used with the VDMS, for predicting mixture density in “air-in-water” flows. This modified model is extended to make use of the predicted coupling stiffness between the VDMS sensing pipe and the multiphase mixture. The modified model gave density measurements with a higher order of accuracy than the earlier model which did not consider the effect of the flow compressibility on the coupling between the sensing pipe and the pipe contents.