Periodic Reporting for period 4 - OpaqueFlows (Flows Unveiled: Multimodal Measurement in Opaque Two-Phase Flows)
Reporting period: 2021-11-01 to 2022-10-31
While we know the exact equations that describe these flows, we cannot solve them analytically. Computer simulations are possible, but demand enormous computational efforts, even for relatively simple problems. The only solution for applied flow studies is therefore a prediction based on simplified models. However, these models lack a solid underpinning, predominantly due to a lack in good experimental data: these flows are difficult to characterize, because they turn opaque even at modest volume loads. Current optical techniques (based on lasers and cameras) will thus no longer work. By making use of non-optical imaging techniques, generally borrowed from a medical setting, we can see in these opaque flows. For instance, imaging based on ultrasound, magnetic resonance, and X-ray will allow us to determine velocities and concentrations – we can clarify these opaque flows.
In this project, we developed and applied non-optical measurement techniques to investigate opaque flows. Three benchmark flows were selected that contain interesting scientific questions, but could also be used to further refine the techniques: particle-laden pipe flow, Taylor-Couette flow, and a cavitating venturi. For each case, we combined the (complementing) data that we obtained from various techniques. This provided an unprecedented insight in each of these flows, so that we now have a better understanding. As some techniques had partial overlap in the data that they provided, we could also refine the measurement methodology. Furthermore, the data will be instrumental in validating numerical simulations, so that industrial applications can reliably be predicted in te future.
For the particle-laden pipe flow, we combined ultrasound velocimetry and magnetic resonance imaging (MRI). The former yielded the velocities of the dispersed phase, while the latter gave velocities of the continuous phase and concentration profiles. Combined, they allowed us to study the transition to turbulence of these flows. We discovered that a different transition scenario occurs at ‘moderate’ volume loads. This fundamental change also explained prior pressure drop measurements: with growing Reynolds number (Re), particle-induced fluctuations lead to turbulence, bypassing the conventional intermittency. The fluctuations increase the friction of the flow at low Re. At higher volume fractions and Re numbers, we observed a decrease in friction. This surprising finding was linked to a change in particle distribution: rather than a homogeneous distribution, the particles concentrate near the center, leaving a lower-viscosity layer near the wall.
In the second benchmark flow, particle-laden Taylor-Couette flow, we investigated the role of particle concentration on the regime transitions (Couette flow to Taylor Vortex flow, etc.). A combination of torque measurement, flow visualisation and ultrasound velocimetry provided detailed insight. Apart from a different critical Reynolds number for the transitions (as compared to single-phase flow), we observed a number of remarkable features. Some of these were very recently reported by others, but some are completely new - e.g. persistent azimuthal structures. As such features are difficult to capture using observations at one location, we devised a ‘360 degrees’ imaging system using 12 cameras. This allowed us to capture (time-resolved) the entire flow domain
In the third benchmark, we investigated a cavitating venturi using high-speed imaging, particle image velocimetry, MRI and x-ray densitometry. Combined, these provided information about the global features, near-wall velocity of the re-entrant jet, the velocity field, and the vapour fraction. This provided a wealth of information not seen before in an experiment. Apart from more insight in the flow, this data is also useful for validation of simulations.
For the measurement techniques, we further refined ultrasound imaging velocimetry. We extended the technique, showing that it can reliably be used to measure turbulent flows using plane-wave imaging. We demonstrated that concentration measurement using ultrasound are possible in theory, but difficulty for moderate volume fractions. For the latter cases, the ‘multiple scattering’ phenomenon inhibits reliable characterisation, unless very elaborate calibration and reconstruction methods are pursued.
By directly comparing x-ray densitometry and conventional imaging, we showed that you cannot infer vapour fractions from imaging data, unless you are in a very dilute regime.
We showed that magnetic resonance velocimetry is a feasible way to obtain velocity information in liquid-gas flows. While we had hoped to also obtain vapour fractions, it turns out that this information is difficult to extract in complex flows.
Regarding the insight in the benchmark flows, significant breakthroughs were achieved in the particle-laden pipe flow case. Apart from discovering a fundamentally new transition scenario, we could characterise the particle-induced fluctuations. By combining information from conventional PIV and ultrasound velocimetry, we could formulate a scaling law that can predict how and when a flow transitions to turbulence, based solely on volume fraction and particle/pipe diameter ratio. Furthermore, MRI gave insight in the distribution of the particles. We found that they tend to concentrate near the pipe center, something not quantified in such great detail before. This piece of information could in turn explain why certain densely-laden flows show a drag decrease.
For the cavitating venturi, a systematic study using various techniques led to a significantly improved understanding. Starting from an understanding based on qualitative information, we could add quantitative information, essential to understand the dynamics of the system. For instance, we now have local volume fractions from x-ray imaging. Furthermore we were able to actually characterise the re-entrant jet using near-wall tomographic particle image velocimetry, rather than rely on estimates from visualisation.