In recent years, emissions of hydrogen cyanide from metallurgical operations have received renewed attention by legislative bodies, leading to the need for a reliable quantification method for HCN volatilisation. Subsequently, the purpose of this project, launched by Anglogold Ashanti Ltd. and in collaboration with MINTEK and the University of Pretoria, was to develop a prediction model for cyanide volatilisation from plant operations and tailings storage facilities in South Africa. The study was done in four stages, the first being a laboratory study of the equilibrium behaviour of hydrogen cyanide. Henry’s Law constant (kH) was determined at different solution cyanide concentrations, salinities and temperatures. A value for kH was established at 0.082 atm.L/mol, which was found to be independent on the solution cyanide concentration between 10 and 200 ppm cyanide. In addition, the effect of temperature on kH was found to be negligible at solution temperatures between 20 and 35ºC. It was also concluded that high salinities increase kH and promote volatilisation, but this effect was negligible at the typical salinity levels found in South African process water. The second stage entailed a detailed study of the mass transfer coefficient, KOL, for hydrogen cyanide from cyanide solutions and pulp mixtures, both in the laboratory and on-site. It followed from this investigation that the most important parameters affecting KOL are the HCN (aq) concentration in the liquid, the wind velocity across the solution or pulp surface, expressed in terms of a Roughness Reynolds number, Re*, and the moisture content, or solid to liquid ratio, of the pulp. Furthermore, it was concluded that KOL is highly sensitive to HCN (aq) concentrations at low concentrations, while it becomes rather insensitive to HCN (aq) at concentrations above 20 ppm HCN (aq) . The data generated by the laboratory and on-site test work was incorporated into the development of an empirical prediction model, based on the Roughness Reynolds number (Re*), moisture content (M), and aqueous cyanide concentration (HCN(aq) ) which may be described by the following equation: KOL= <font face="symbol">a</font> Re*b Mc HCN(aq)d + e The model coefficients were subsequently determined for application of the model to leach tanks, adsorption tanks, tailing storage facility surfaces and return water dams. The calculated model predictions for KOL were in excellent agreement with the measured test work data. Finally, the prediction model was validated at the leach and adsorption sections of a selected gold plant and a selected tailings storage facility. The model predicted that 9% of the cyanide lost in the leach and adsorption section could be attributed to HCN volatilisation. As for the tailings storage facility, the model assigned 63% of the cyanide lost from the tailings storage facility to HCN volatilisation, of which 95% occurred from the area on the tailings dam surface covered in a thin liquid film. It is recommended that the current methods available for the determination of HCN (aq) be further improved, due to the sensitivity of the model to the input value of the HCN (aq) concentration, in order to ensure that reliable predictions are made. It is also suggested that additional validation work be done in order to establish the generic applicability of the model to different sites.
Dissertation (MEng(Metallurgical))--University of Pretoria, 2007.