Visualizing Flow in Three Dimensions Three Ways
For designers of chemical processing equipment, scaling up from prototype to production is difficult because some mechanical characteristics vary greatly when you change their size. For example, mechanical extractors used to mix and separate fluids are made in various sizes, but the way in which liquids churn through them can change with the scale.
According to Jyeshtharaj B. Joshi, director of the Institute of Chemical Technology at the University of Mumbai in India, experimental data from a prototype often is used to devise production systems. For example, the separation capacity and the efficiency of a prototype extractor are used to design a larger system. Unfortunately, the way in which larger extractor sizes affect flow patterns is not well understood.
“This is mainly because of complexity of fluid dynamics and lack of our understanding regarding the relationship between the fluid mechanics and the design objective,” Joshi said.
Joshi, along with Sandesh S. Deshmukh at the university and Sudhir B. Koganti of the Indira Gandhi Centre for Atomic Research in Kalpakkam, India, have been working on solving those unknowns. They have been investigating annular centrifugal extractors, devices in which an inner cylinder rotates with respect to an outer stationary cylinder. This type of extractor mixes one liquid into another and then quickly separates them into heavy and light phases.
Such extractors are commercially important and have been in use for 30 years. The flow patterns within them have been studied for nearly as long. The investigators’ method for studying these patterns involves the simultaneous use of simulations based on computational fluid dynamics and on flow visualization based on laser Doppler velocimetry and particle image velocimetry.
The use of two velocimetry techniques improved confidence in the measurements because one served as a check on the other. Laser Doppler velocimetry is a point measurement technique, whereas particle image velocimetry captures readings across a plane. “Particle image velocimetry gives a better feel for the results, as it gives a complete picture of the flow in the domain. Laser Doppler velocimetry, on the other hand, gives high data rate and is useful for time-dependent properties,” Joshi said.
As described in a paper released online Dec. 11, 2007, by Industrial & Engineering Chemistry Research, the scientists constructed an annular centrifugal extractor with transparent acrylic and used a solution of sodium iodide in water to match its refractive index. They seeded the solution with 20-μm-diameter silver-coated hollow glass particles for imaging. For measurements, they used a particle image velocimetry system from TSI Inc. of Shoreview, Minn., and a laser Doppler velocimeter from Dantec Dynamics A/S of Skovlunde, Denmark.
An annular centrifugal extractor consists of a rotating cylinder inside a stationary one as well as various mixing vanes, or baffles. The extractors are used to mix and then quickly separate liquids into heavy and light phases, but building up from a small prototype to a large production unit with confidence is difficult because of the complex nature of the flow of the liquids. Images reprinted with permission from Industrial & Engineering Chemistry Research.
The particle image velocimetry system’s light source was an Nd:YAG laser from New Wave Research Inc. of Fremont, Calif., set to fire 6-ns pulses shaped by a combination of lenses into a 1-mm-thick sheet that spanned the sample fluid. A 4-megapixel CCD camera mounted perpendicular to the sheet captured the images, with the flow determined by the movement of the particles in consecutive shots.
For laser Doppler velocimetry measurements, the setup consisted of three mutually perpendicular beams derived from a Spectra-Physics argon-ion laser. The wavelength of one beam was 588 nm; the second, 614 nm; the third, a combination of the other two. Photomultiplier tubes captured the laser light scattered by the seeded particles in the solution. The system determined a point-source-velocity profile from frequency shifts in the light.
Shown here are a vector plot of liquid flow at extractor rotational speeds determined by computational fluid dynamics (left) and the actual results obtained with particle image velocimetry (right).
The researchers performed computational analysis by dividing the reactor volume into small elements using a 3-D grid. They solved equations regarding continuity and motion repeatedly for each element. Although computational fluid dynamics has been used for decades, it is now more feasible to apply the method to such problems, Joshi said. “The utility of computational fluid dynamics has been improving as the computation power increases over the years.”
The velocimetry data validated the computational model which, in turn, allowed the researchers to look at the effect of start-up procedures and rotation speeds, flow patterns in the presence of axial flow and the effect of radial baffles in the annulus on flow patterns. Based on these results, Joshi noted that moving toward a plug flow in the reactor would minimize the reactor’s volume, a desired result that soon could find applications outside of the laboratory.
“This product is expected to be commercial in less than 18 months,” he said. He added that an optimized design could be used in setups with an output of as low as a few liters up to 50 tons per hour.
Contact: Jyeshtharaj B. Joshi, University of Mumbai; e-mail: email@example.com.
MORE FROM PHOTONICS MEDIA