97 rate still further This effect has also been noted by Bird

The onset of vigorous splashing has been associated with an increase

Q/- QA YQ 0*|

collected droplets ejected from a bath of water by an impinging CO^ jet and has noted that their CO^ content can be three to six times greater than that in the bulk bath.

Demin has used the transfer of ammonia to a bath of water, con­ taining phenol phthalein indicator to assess the efficiency of mixing produced by an impinging ammonia-air jet. Increased mixing efficiency was shown to be associated with increased jet penetration and also by the use of an inclined nozzle.

Heat transfer analogies have been used to simulate mass transfer 8S 86

from a gas jet to a liquid bath. Sheridan and co-workers ' ’ have used a model involving the cooling of a water bath from rJO°G to 40°C

93

by an impinging air jet. Huang-Kim-Ho has used a heat transfer analogy to simulate mass transfer when oxygen lancing an open hearth furnace. In this case an air jet, heated to 140-150°C, was introduced into a bath of aqueous potassium iodide solution, covered by a layer of oil to sim­ ulate the slag.

Chatterjee and Wakelin have examined the transfer of oxygen from an oxygen jet to a pool of molten silver, under conditions such that splashing did not occur. They we re able to determine the mass transfer coefficient for the process and showedthat the majority of oxygen trans­ fer was occurring in the vicinity of the impact crater^*

A number of workers have attempted to produce a more complete model of the L.D. process by involving a reaction between the jet ge,ses

95

and a solute in the liquid bath which produced a second gas. Van Langen was one of the first to do this by impinging a jet of hydrogen chloride onto an aqueous solution of sodium, bicarbonate, on which a paraffin layer floated to simulate the slag. The hydrogen chloride from

and reacted with the carbonate in solution to produce carbon dioxide , Na H CO.. + HC1 NaCl + Ho0 + C0o f

J £ £

94 90

Similar systems have been used by Andonev and Shimotsuma •

In an attempt to simulate sulphur and phosphorus transfer from the metal to the slag, Andonev dissolved potassium iodide and iodate in the aqueous solution. In the presence of hydrochloric acid, iodine was formed which was transferred to the paraffin slag.

6HC1 + 5KI + KIO^ -*• 3H20 + 6KC1 + 3Ig

For similar reasons. Shimotsuma in his model attempted to monitor the transfer of methyl orange from the paraffin slag to the aqueous phase.

Whilst the above models involve a reaction between the gas jet and the bath there is no reaction between the bath and the slag, since the paraffin layer is effectively inert. This is a fairly serious criticism in view of the importance placed upon the formation of foams and emulsions in current theories of the mechanism of L.D. refining.

2.3 SINGLE DROPLETS

Extensive reviews of the literature concerning the behaviour of

102 103

single droplets have already been made by Crimes and Aeron and hence only relevant features, with respect to the present work, will be outlined.

2.3.1 Hydrodynamic behaviour of falling droplets

A drop of one liquid falling through another behaves as a rigid sphere and obeys Stokes Law only when the Reynolds number is much less than unity. For steel in slag this corresponds to a diameter of less than 0.5 mm. As the diameter increases, internal circulation starts to

i

occur within the drop, caused by the viscous forces operating at the

105

surface with the continuous phase . Internal circulation tends to produce an increase in velocity beyond that predicted by Stokes Law. As the size increases further, the drop begins to deform to an oblate spheroid and the velocity eventually reaches a maximum value. The peak velocity is generally associated with increased droplet distortion and the onset of oscillations, both of which tend to reduce the velocity. Ultimately a size is reached at which the oscillations are so pronounced that the surface tension forces are no longer sufficient to prevent the break up of the droplet.

The generalised curve for the terminal velocity of droplets, falling through a liquid phase, as a function of droplet size is illustrated

in Figure 5*

A number of workers have attempted to produce theoretical and empirical corrections to Stokes Law, allowing for various modes of

circulation within the drop. These only tend to he useful at low Reynolds numbers.In metallurgical processes, the droplets of interest have diameters between 1 and 5 mm with Reynolds numbers between 100 and 1000 r^ggQ have both a wake from which vortices are t o m off and well developed internal circulation patterns.

The most successful relationship between the terminal velocity of 107 a free falling droplet and its size is that attributed to Hu and Kintner They determined the terminal velocities of droplets of ten organic liquids, falling through a stationary water phase, covering the Reynold number

range 10 to 2200. A correlation was obtained, for nine of the ten systems,