FFF - Ferrolegeringsindustriens Forskningsforening

PLASMA PROCESS TECHNOLOGY.

The project was initiated to obtain competence on plasma technology and to prepare the Norwegian ferroalloys industry to take in use new plasma based proecss. The Plasma Reactor for Smelting at SINTEF (PRESS) was designed and built. Test runs including remelting of silicon metal fines, production of ultrafine silica and a novel three step process for production of silicon metal were carried out. In parallel to the actual experiments, theoretical work on numerical simulation of DC arcs in a reactor enviornment was done. Equipment and know-how built up during this project have given birth to an extensive industrial plasma research activity at SINTEF/NTNU.

The plasma reactor was built with flexibility in mind, and is normally equipped with three plasma torches in the transferred or non-transferred arc mode.The DC supply includes three spearate power units with current rating 500 A and open circuit DC voltage 300 V. Two or three units can also easily be connected in oarallel or series to give a maximum current at 1500 A or a maximum voltage at 900 V. The reactor, which is usually operated with graphite inner lining, consists of several modules which can be modified or replaced to suit the process requirements.

The three plasma torches are inclined 45o and symmetrically positioned in the conically shaped top module. A single vertically mounted plasma torch has also been used. The torches are supplied by Plasma Systems Ltd., England, and are of standard design with a pencil shaped 2% thoriated tungsten cathode and a copper nozzle/anode. The current rating is 600 A DC.

A control room contains all the equipment necessary to control the power supplied to the plasma torches and the raw material feed rates. It also contains instruments that show important process parameters. Data logging equipment is used to survey and monitor each test run. The cooling water system is divided into several independent circuits. The optical unit provides visual observation on a screen or a TV monitor as well as video recording of the interior of the reactor during operation.

The PRESS reactor concept offers advantages such as short response time, excellent process control and high production per unit volume. The coupling between electric resistance and the raw material properties is weak compared with submerged arc furnaces. The reactor is well suited for various applications, and scaling up to larger industrial installations should be fully feasible.

Numerical modelling fo DC arc discharges - ordinary free-burning as well as transferred plasma arcs - in a metallurgical reactor enviornment has been an important activity. Two problem areas of great practical importance are emphasized:

• Heat exchange by convection and radiation between the arcs and the reactor surfaces.
• Mass and momentum transfer across the metal-plasma interphase including mass diffusion in the arc region.
  

3D models have been made of SINTEF's experimental plasma reactor PRESS. The geometry of this reactor bears a certain resemblance to more conventional 3-phase arc furnaces. The thermally isolated reactor lining and the molten metal pool are included in the computational domain. In addition to arc-to-wall radiation the important wall-to-wall radiation is also taken into account. Thus the heat flux distribution as well as the surface tempratures of the reactor walls are computed.
 

In order to verify the simulations, measurements were done in a calorimetric 1:1 cold-wall model of the PRESS reactor. The heat distribution on the various wall sections and the metal pool was determined with different arc lengths. The calorimetric model is equipped with the same olasma torches and operated at the same power level as the PRESS reactor. Although the current distribution in the 3D simulations was "prescribed", the agreement with the calorimetrically measured heat distribution is satisfactiry.

The predicted over-all thermal efficiency compared well with efficiency measured during practical tests with remelting of silicon metal fines in the PRESS reactor. Further improvements are expected if the current distribution is computed - rather than "prescribed" - by solving Maxwell's electro-magnetic equations. These are implemented in our 2D arc models.

In many industrial applications one of the two arc attatchment spots is situated on the surface of the liquid metal to be produced or processed. The arc region will then be infiltrated by metal vapour, which must be expected to influence arc resistance, arc radiation, heat transfer to the metal pool, etc. In addition, the arc(s) will generate stirring in the metal pool. Example: a transferred DC arc burning in argon against an anodic silicon metal pool. Computer codes were made to calculate chemical composition, thermodynamic data, transport coefficients and radiation properties of Ar-Si plasmas.This work has been extended to 4-component Si-O-C-Ar plasmas. When modelling the crater zone including the AC arc in industrial submerged arc furnaces for silicon metal production, data for SiO-CO mixtures are required.

The complex interaction between the arc and its evaporating anodic metal pool has been simulated by employing two coupled sub-models:
i)    The plasma phase model, where the chemical diffusion equation is solved together with the mass, momentum and energy conservation equations and Maxwell's electromagnetic equations.
ii)   The metal pool model, where the velocity and temperature are calculated taking into account electromagnetic forces as well as surface shear due to cathode jet momentum and surface tension gradients.

The following figures show the temperature and silicon vapour distributions and the stream line patterns in a plasma reactor with a 1000 A, 130 mm long transferred DC arc burning between a plasma torch and an anodic silicon metal pool.