Batch Consistency with On-Line Particle Fineness

Using Insitec to measure on-line particle fineness, final grade cement is now monitored continuously every 30 seconds at Ashgrove Cement, which leads to several kilograms being measured over the course of 4 hours. The Insitec PC is connected to the plant distributed control system via 4-20mA current loops which report the following information used by Ashgrove to control fineness of the cement:

• mass fraction (%) below 30µm
• mass fraction (%) below 44µm
• mass fraction (%) between 8 and 23µm
• "calculated" Blaine ie the specific surface area reported by Insitec was correlated against manual Blaine measurements
• 10-minute average %<30µm for a PID loop to automatically control the separator speed
• laser transmission (%) (which correlates to loading in the process stream)

Soon after the Insitec installation, 30 Type I cement samples were analyzed off-line with Insitec. Initial correlation between Insitec and measured Blaine number was poor. The primary limitation was that the span of the data population was only 2 to 3 times the expected error of the measurement. To mitigate this problem three data sets have been combined using 30 samples from 2 different plants of the same company The combined data set (Figure 3) has a span of 700 cm²/g (70m²/kg) and, as expected, the correlation coefficient R2 values improved from a value of 0.43 to 0.79. The standard error is 7.15 m²/kg (71 cm²/g) or 1.8% of the measurement. This is slightly more than the ASTM estimated error of 1.5% for the air permeability test using the Blaine apparatus. The error bars span +/- 7.15m²/kg. The correlation is given in the figure below:

Figure 1: Correlation between off-line Blaine measurement and on-line Insitec SSA measurement

The figure below shows a 12 day comparison between the Insitec SSA (Specific Surface Area) and the traditional Blaine number measured by pressure drop. Note the slow equilibration to steady-state operation of the process after a 1 day shutdown.

Figure 2: Comparison of Insitec SSA with traditional Blaine measurement during 12 days of production.

Erroneous off-line measurements

The use of on-line analysis has resulted in changes to accepted practice at Ashgrove Cement. An example of this occurred whilst milling clinker sourced from external storage. "Outside" clinker is treated with respect as it has already commenced the hydration process and consequently has a high ignition loss. Past practice of the control operator was to slow the separator as the manually measured Blaine value increased. The online analyser over a similar period (graph below) clearly shows that the separator speed should remain unchanged as the actual particle size variation is minimal.

Figure 3: Graph showing erroneous Blaine result

It can be seen that even though mill conditions are steady the manual Blaine gave an erroneous result. If the operator were still in control an adjustment to separator speed would have been in order at that point.