XXXI International Mineral Processing Congress 2024 Proceedings/Washington, DC/Sep 29–Oct 3 1089
partly filled with more rigid pre-compacted material. In
reference to Johanson’s compressibility factor, higher values
will be achieved when examining pre-compacted material.
An example in potash shows that this value may be 48%
higher than the compressibility factor of the same material
prior to being compacted (fresh feed).
Therefore, an irregular roller movement and oscilla-
tions in the working gap typically accompanies the material
glazing, and as a result, the pressure of the hydraulic sys-
tem also will oscillate. Furthermore, it leads to a significant
drop in the throughput of the roller press and thus product
quantity. Typical sensor readings for both gap measurement
and hydraulic pressure can be seen in Figure 10.
The Figure shows that oscillations in both the gap mea-
surement and pressure will increase over time. A difference
in operating gap, between the drive and non-drive side is
noticed due to this material build up. This knowledge can
be used to build a digital helper for early detection of glaz-
ing conditions. This helper acts as a soft sensor, which uses
a combination of data from different physical sensors to
approximate variables which themselves cannot be directly
measured using a physical sensor alone. Since oscillations in
gap size can also be seen in startup- or shutdown-processes,
a filter was applied to only process stationary produc-
tion data. It was discovered in experimental investiga-
tions that temperature-driven glazing can be counteracted
by changing the compressive force. With this additional
knowledge, a control recommendation can be created to
change compacting pressure when detecting glazing on the
roller surface.
Wear detection in High-Pressure Grinding Rolls
Another type of soft sensor is applied for wear detection in
high-pressure grinding rolls. The roller surfaces of HPGRs
are subject to different levels of wear depending on feed
material characteristics and process settings (Hanstein
2002). With time, the length of the embedded studs on the
surface will decrease. Typically, the wear rate is estimated by
manually measuring stud length and the remaining lifespan
is calculated using linear approximation.
During operation, the roller gap is kept constant by
controlling the hydraulic and feed systems. This means that
a decrease in stud length will not be observable by measur-
ing gap distance, as the gap distance is measured from the
bearing block. At the same time, this points to the mea-
surement of absolute position of the floating roller as the
overall decrease in diameter will result in a change in roller
position versus a fixed point, for example on the roller press
frame. The operating data is stored on an edge computer,
where algorithms manage the calculations for the predicted
life expectancy. This information can then be forwarded to
machine specialists via IIoT remote analysis services, and
Figure 10. Example of gap settings and hydraulic pressure while glazing is occurring (at 22:30)
partly filled with more rigid pre-compacted material. In
reference to Johanson’s compressibility factor, higher values
will be achieved when examining pre-compacted material.
An example in potash shows that this value may be 48%
higher than the compressibility factor of the same material
prior to being compacted (fresh feed).
Therefore, an irregular roller movement and oscilla-
tions in the working gap typically accompanies the material
glazing, and as a result, the pressure of the hydraulic sys-
tem also will oscillate. Furthermore, it leads to a significant
drop in the throughput of the roller press and thus product
quantity. Typical sensor readings for both gap measurement
and hydraulic pressure can be seen in Figure 10.
The Figure shows that oscillations in both the gap mea-
surement and pressure will increase over time. A difference
in operating gap, between the drive and non-drive side is
noticed due to this material build up. This knowledge can
be used to build a digital helper for early detection of glaz-
ing conditions. This helper acts as a soft sensor, which uses
a combination of data from different physical sensors to
approximate variables which themselves cannot be directly
measured using a physical sensor alone. Since oscillations in
gap size can also be seen in startup- or shutdown-processes,
a filter was applied to only process stationary produc-
tion data. It was discovered in experimental investiga-
tions that temperature-driven glazing can be counteracted
by changing the compressive force. With this additional
knowledge, a control recommendation can be created to
change compacting pressure when detecting glazing on the
roller surface.
Wear detection in High-Pressure Grinding Rolls
Another type of soft sensor is applied for wear detection in
high-pressure grinding rolls. The roller surfaces of HPGRs
are subject to different levels of wear depending on feed
material characteristics and process settings (Hanstein
2002). With time, the length of the embedded studs on the
surface will decrease. Typically, the wear rate is estimated by
manually measuring stud length and the remaining lifespan
is calculated using linear approximation.
During operation, the roller gap is kept constant by
controlling the hydraulic and feed systems. This means that
a decrease in stud length will not be observable by measur-
ing gap distance, as the gap distance is measured from the
bearing block. At the same time, this points to the mea-
surement of absolute position of the floating roller as the
overall decrease in diameter will result in a change in roller
position versus a fixed point, for example on the roller press
frame. The operating data is stored on an edge computer,
where algorithms manage the calculations for the predicted
life expectancy. This information can then be forwarded to
machine specialists via IIoT remote analysis services, and
Figure 10. Example of gap settings and hydraulic pressure while glazing is occurring (at 22:30)