Title:

Computer simulation of the atmospheric arc plasmas environment for production of nanoscale materials

The aim of this study is concerned with the development of computational models for two types of atmospheric technological plasmas with background applications in material processing (waste disposal and production of nanoscale powders) and formation of nanoscale carbon structures (formation of carbon nanotubes and fullerenes). . The twin torch system consists of two electrode assemblies, each of which has a central electrode surrounded by a nozzle to confine the shielding gas. The two electrodes are configured to have an angle between their axes. The arcing environment is truly three dimensional (3D) and thus requires a three dimensional model to simulate the arcing process. Representation of 3D shape in the computational domain is a challenging issue and needs to be addressed before a sensible solution is obtained. In the present model a novel numerical scheme is developed to conveniently and automatically allocate cells (each cell is a finite volume in the solution procedure) to a 3D shape. This has greatly enhanced the adaptability of the model to cope with complex geometries encountered in arc plasma systems. One of the key issues in 3D arc modelling is the correct calculation and application of the Lorentz force terms in the momentum equation. For a symmetric arcing case, such as a free burning arc between a conic cathode and a plate anode, slight asymmetry in the distribution of Lorentz force can substantially affect the arc flow hence resulting in a nonsymmetric arc column. Validation of the numerical method has been carried out using the free burning arc case by comparing the model predictions with measurements and other available simulation results. It has been shown that the present 3D model produces satisfactory results. Since the two electrodes of the twin plasma system are water cooled, erosion of the electrode material is minimal and can be neglected. It has been found that the coupling of the two jets is through a thin, tissue like, conducting layer between the two jets. The cross sectional shape of the two jets is deformed by the Lorentz force. The jets are never completely merged. The Lorentz force induced by the arc current and the combined magnetic field of the two jets tends to move the point of separation away from the electrode tips. It is the strong Ohmic heating, resulting from the high temperature (high electrical conductivity) and high electric field in the coupling zone, that heats up the incoming cold gas to a conducting temperature and thus stabilises the electric current path and subsequently the point of separation of the two jets. It is also shown that the change in flow rate does not have significant influence on the parameters of the plasma. The predicted arc voltage is found to be higher than the measurement. The difference is partly attributed to the statistically high frequency fluctuation of the jet coupling zone. The use of a simple turbulence model brings the prediction closer to the measurement at low current, however results in an excessive reduction in arc voltage at high current. Therefore the use of a conventional turbulence model with a fixed turbulence parameter c to simulate the effect of fluctuations of the jet coupling zone may not be appropriate. The carbon arc simulated in Chapter S is confined by liquid water and bums in the gap between a conic anode and a plate cathode. Such an arrangement aims at maximising the anode erosion rate for the formation of solid carbon Ã‚Â·structures (carbon nanotubes and fullerenes). A mathematical model has been developed to simulate the dynamics of bubble growth and development of the ~c column. A novel numerical scheme is employed to detennine the speed of bubble surface expansion. To simulate the removal of gas from within the domain associated with bubble detachment from the discharge zone, an exit is imposed on the bubble surface when its radius reaches 9mm. Results show that there is large pressure change within the first half millisecond of the discharge. The pressure can be as high as 38bar. Under such a high pressure the water surrounding the bubble is accelerated which leads to an increasing speed of bubble growth. The rapid growth of bubble size under its inertia caused the pressure inside the bubble to rapidly decrease. The arc voltage, which is an indication of the total electrical power input into the domain, settles down in the first millisecond and then maintains a value of SV. The arc temperature at initiation is 24000K within a thin hot column~ This temperature is quickly reduced to l1000K in SJ.lS as a result of radiation loss from the arc column. The axis temperature inside the arc column at 30ms is 8S00K which is higher than the experimental value measured using a spectroscopic method. The possible reason for this difference could be the instability of the arc column which makes the measured temperature as an average value of the arc column. There exists a relatively large area around the arc core whose temperature is within the range of4S00K to 7000K which is the temperature range from measurement. A quasisteady state of arcing is reached within 30ms of arc initiation. The loss of carbon vapour from the computational domain resulting from formation of solid carbon matters is considered by a flux controlled model. Results show that the major loss of the carbon species is at the arc edge which accounts for 99% of the total carbon loss. The predicted arc column voltage is SV which is different from the measurement of 16V. The difference is contributed to the cathode voltage fall which is in the order of the ionisation energy of carbon atoms (11.26eV). Taking this aspect into consideration, the prediction falls into the expected range. The biggest difference between the prediction and measurement is the water evaporation rate. The arc model prediction is only a very small fraction of the measurement. However, it is shown that the measured rate of water evaporation may be incorrect since even when the total electrical power input at the measured arc voltage is used, the amount of water vapour that can be created is only 12.5% of the measured value. This renders the water evaporation rate comparison invalid and further measurements are required. Future work is discussed at the end of the thesis. Three aspects have been identified that need to be addressed. The instability of the twin torch jet coupling may need to be simulated by a transient model. The cathode sheath is important in determining the whole arc voltage and therefore needs to be simulated using kinetic theory. The effectiveness of three dimensional modelling suffers enormously from the prohibitively long computing time because of the strong reiaxation that has to be used to obtain converging results. This naturally requires parallel processing given that high processing power PCs (2.lGHz Pentium Core Duo) have already been used for the present work.
