Gibbs Energy Minimisation for Process Simulations an Example in Uranium Solvent Extraction

Abstract

Mass and energy balance simulations form an integral part of metallurgical process design, from concept level studies through to detailed design. To build an efficient process plant, that maximises production with least capital and operating costs, an accurate model of the flows and the chemistry involved in the process is indispensable for all but the simplest of processes. For existing operations, a mass and energy balance is essential for optimisation of the existing process. Also, the consequences of proposed modifications to the process can be assessed quickly, before purchasing any new equipment. In an engineering study, a mass and energy balance is usually created at the concept level, with limited detail, and expanded in the later study phases to contain more detail of both the chemistry and the minor process flows. The selection of tank and pond volumes, pipe diameters, materials of construction, power requirement for motors, and most other design parameters rely on accurate predictions of the conditions that the plant will encounter through the life of the resource.

A mass and energy balance model, as with all models, will always be a simplified representation of reality. The level of detail with which a process is simulated need only be enough for the purpose for which it is being created. As an example, the chemical phenomena occurring in the high temperature and pressure soup of an HPAL autoclave need not be modelled unless they are relevant to the design of the autoclave or downstream processes. The precise reaction or mechanism is usually irrelevant in this case, provided the distribution between phases and the oxidation states of the elements is adequately characterised. While valid in many instances, this philosophy is an easy one to apply irresponsibly. When forming a process design criteria, previous experience from engineers will often be agreed upon as the basis for critical design parameters, but these figures may be biased towards similar, but not identical processes, where a critical factor has not been considered that may drastically affect the result. Future phases of work, and testwork, should hopefully decrease the risk associated with this, but in some cases, the outcome could be disastrous. In this case, had the model been more predictive and less biased towards past experiences, alarms warranting investigation may have been triggered much earlier in the process design.