10th International Aerosol Conference September 2 - September 7, 2018 America's Center Convention Complex St. Louis, Missouri, USA
Abstract View
Uniformity of Particle Concentration after Mixing Aerosol Flows
David Walker, Tyler J. Johnson, Robert Nishida, JONATHAN SYMONDS, Kingsley Reavell, Cambustion
Abstract Number: 1351 Working Group: Instrumentation
Abstract Aerosol experiments are frequently performed with low flow Reynolds numbers (Re<4000), and it is often necessary to mix two streams (for example an aerosol flow with clean make-up air), then later divide the mixed stream between different instruments. It is important that the combined stream is uniformly mixed before the division occurs, otherwise the concentration in the two substreams may be significantly different.
In this work, the variation in aerosol concentration across the cross section of a tube carrying a flow from different designs of tee mixing an aerosol flow from a nebuliser with filtered dilution air was measured by sampling from the flow through a 0.7mm capillary tube. The range of flow rates tested and the velocity profile in the sample tubes prevents true isokinetic sampling by this capillary so the experiments were performed with a NaCl aerosol of approximately 84 nm CMD (mobility diameter) to minimise artefacts from this sampling.
Two 6mm diameter tees of different geometries were tested with a 0.8 l/min aerosol + 1.2 l/min diluent flow. For a plain tee, the concentration distribution measured 30 tube diameters downstream of the mixing axis showed a 38% total concentration variation (ie. minimum to maximum) of the mean. A coaxial geometry, with the sample flow introduced into the centre of the dilution flow via a coaxial 6mm tube gave a concentration variation of 36%, not significantly different from the plain tee result.
With the length of the downstream tube increased to 400 tube diameters, the variation was reduced to 17%. Introducing a 14 element static mixer produced a 2.4% variation at the 30 tube diameter location, roughly comparable with the uncertainly of the experiment. The static mixer was additionally tested for particle losses which were found to be less than 5%.
The 6mm coaxial tee and a 10mm version were tested at higher flow rates. For the 6mm tee at 30 tube diameters, using 1.7 l/min sample and 5 l/min dilution flow, the variation was 35%; while using a 2.5 l/min sample and 20 l/min dilution flow, the variation reduced to 3.0%. The 2 l/min, 6.7 l/min and 22.5 l/min conditions correspond to Reynolds numbers of approximately 460, 2200 and 7700 respectively.
The 10mm tee measured at 30 tube diameters with 22 l/min flow produced a variation of 4.7%. Closer to the mixing point, at 6 tube diameters, the variation was higher at 15%. Flows of 32 l/min and 62 l/min produced variations of 17% and 27% respectively at that location. Reynolds numbers for the 22 l/min, 32 l/min and 62 l/min conditions were approximately 3700, 5400 and 10500 respectively.
These results suggest that mixing at typically laminar Reynolds numbers (< approximately 4000) even with long tube lengths downstream of the mixing tee is poor and that a mixing element should be used. For low flows, of the order of a litre per minute or less, it is difficult to achieve higher Reynolds numbers. For experiments performed at higher flows where turbulent Reynolds numbers can be achieved, satisfactory mixing may be achieved if tube lengths are sufficient: higher turbulent Reynolds numbers will require longer mixing lengths.