X-ray fluorescence (XRF) is used routinely in cement and clinker production for quality management at cement plants. The gold standard for the cement industry has traditionally been wave length dispersive XRF (WDXRF) for analysis. However, in recent years, energy dispersive XRF (EDXRF) has matured to the point that it is now an alternative technique for routine laboratory analysis. Both techniques can use either pressed pellets or fused beads as the final sample for analysis. The preferred choice between these two techniques depends on user requirements and, typically, pressed powder pellets are suitable for process control, while glass beads produce the most accurate results.
As a complement to laboratory analysis, high speed process control systems can use on-line or at-line analysis to make continuous high frequency adjustments in order to reduce composition variability. Today, many plants already benefit from the on-line analysis of raw material fed to the raw mill, using prompt gamma neutron activation analysis (PGNAA) to determine the material composition. Software is used to adjust the mixture composition to the desired target. Frequently, the software is operated with synchronised inputs from laboratory analysis, which is an outer loop for bias correction of the on-line analyser (the inner loop), enabling a more accurate control action.
An alternative to on-line analysis on a conveyor belt is post raw mill at-line analysis, where the analyser is continuously monitoring high frequency samples from the process. An example of this system is the FCT ACTech RMX, which can be used for the powder analysis of raw meal or cement.
Sampling and analysis time
On-line and at-line analysis minimises three issues with laboratory analysis: sampling, preparation, and analysis time. It is important to recognise that, despite the relative accuracy and precision of all combinations of laboratory techniques, the most likely differences between the analysis and the process are those introduced by sampling and preparation errors. This is fundamental to the problem of attempting to provide an analysis of a micron’s thick surface on top of a few grams of sample, which represents perhaps a few hundred tonnes of material.
In addition, with laboratory analysis there is a significant delay between collecting the sample and what has happened in the process when the analysis becomes available. This delay can be several hours and during this time many hundreds of tonnes of off-target mixture can be introduced to the process. The problem of delayed analysis is obvious, but sampling errors can be subtler. To collect the sample, several aliquots are extracted from the process at appropriately spaced intervals. When there is a periodic variation in the process, it becomes important that the timing of aliquot collection does not correlate, otherwise the final sample may be biased high or low, depending on whether the timing matches a peak or trough in the variation. Clearly, in this example, one sample will be biased high and the other sample will be biased low.
To read the full article, claim your FREE TRIAL of World Cement here
Read the article online at: https://www.worldcement.com/special-reports/03062019/x-ray-vision/