Coal Preparation » Fine Coal
Jameson cells are often used in a coal processing plant to treat the ultrafine coal (-250μm) and reduce the ash content of the final product. However, it is not uncommon for these cells to be poorly controlled leading to unnecessary loss of combustibles to the flotation tailing stream. This is because the Jameson cell often employs a stabilising control strategy where the major operating variables are controlled to a pre-determined set point. This type of control system does not consider the effect of the disturbances on the flotation process and thus controls for process stability instead of optimising performance. There is an opportunity to improve the flotation process by shifting from a stabilising control strategy to an optimising control strategy. A commonly used algorithm for optimising control is model informed predictive control (MIPC) which is proactive, i.e. it predicts the effects of a change in process variables on flotation performance in real time and then provides an appropriate control action. For the development of a MIPC strategy, it is important to understand the relationships between process variables (i.e. operating conditions and feed characteristics), on the flotation performance. It is also important to be able to measure continuously and in real time the key variables that input into these relationships.
The objectives of this project included developing empirical models to relate the process variables to the flotation performance as well as identifying easily measured proxies that can be used to infer the flotation performance in real time. The overall objective being to suggest a model informed control strategy for coal flotation in a Jameson cell.
To achieve the objectives, two experimental campaigns were conducted on a coal preparation plant. During the campaigns, supplied air rate was altered while collecting the information required to calculate the flotation performance and the parameters to determine the performance proxies. A variety of different sensors for measuring information continuously were tested. Two coal types sourced from two different seams were being treated during the experiments.
The data collected were used to develop statistical empirical models that correlated combustible recovery, ash rejection and product ash content to superficial gas velocity, froth depth, feed solid content, coal type and/or feed ash content. Using these models, it was found that a maximum in combustible recovery as a function of superficial gas velocity exists. This superficial gas velocity value at which the maximum combustible recovery was attained was independent of froth depth, coal type and feed particle size, but was dependent on feed percent solids. A potential control strategy to maximise combustible recovery is to operate at the superficial gas velocity at which the peak is attained. The position of the peak can be estimated if the feed percent solids is known.
Product ash content was found to increase with superficial gas velocity. This control strategy would therefore benefit from a prediction of product ash, enabling the superficial gas to be constrained to one that maximises combustible recovery whilst not exceeding a product ash target. Product ash in the experimental program was able to be predicted reasonably accurately as a function of superficial gas velocity, froth depth, feed percent solids and feed ash, however for this prediction to be robust, it would need to be developed from an experimental program involving testing of a greater range of process variables.
The best proxy of performance was found to be froth velocity. It was found to be moderately correlated with the solids concentrate flowrate.