Investigation of Feed Dynamics Using the Radiotracer Residence Time Distribution Method and Metastable Techenetium-99: a Case Study of Clinker Milling at National Cement Company, Athi River

Abstract

In Kenya, cement is a key ingredient in the built and construction industry and in 2017, its production contributed to 5.8% of GDP. In general, cement production is energy intensive and costly. Currently, National Cement Company Limited (NCCL) is the second largest cement producer in Kenya, with an annual production capacity of over 2 million. In 2018, mill 1, its oldest clinker grinding ball mill at their Athi River plant, was suspected to malfunction due to an increase in its power consumption. The mean residence time (MRT) and material hold-up in the mill were also not known. This study, therefore, assessed the flow of porous clinker through the mill to determine if there were any deviations from previously observed models that were affecting its efficiency. This was done using the radiotracer technetium-99m (Tc-99m). Radiotracers are preferred over conventional tracers since they enable online measurement, thereby allowing the system to be studied without interfering with the normal operation of the mill. The radiotracer residence time distribution (RTD) investigation was carried out by introducing clinker labelled with the Tc-99m at the inlet of the mill. Its activity was then monitored at the mill inlet, the mill outlet and at the separator recycle line, using three NaI(Tl) scintillation detectors at each location. The data recorded by the detectors was used to plot the RTD curves, to compute the mean residence time (MRTs) and material hold-up within the clinker ball mill and its separator. The MRT of material within the mill was found to be 980.5 s, whereas the material hold-up was 13.6 tonnes. Further, the empirical RTDs were modelled in DTS PRO software to characterize the flow of material within the ball mill and separator. The results indicated that the flow could be described using two models: an axial dispersion model (ADM) and a tank-in-series model both connected in series to a plug flow component. The two models revealed a significant degree of axial mixing/back-mixing in the ball mill, as indicated by the low peclet number in the axial dispersion model, and the low value of the number of tanks in the tank-in-series model; this meant that dispersive effects were predominant in the mill resulting in inefficient operation. There were, however, no other flow anomalies observed from the modelling. It is recommended that the operation and design parameters of the ball mill be adjusted to eliminate the observed back-mixing.