Anode used in an aluminium smelter is made up of a precise combination of coke and coal tar. Coke should be of a specific size range to achieve the appropriate strength in the anode. Ball Mill in a Green Anode Plant (GAP) grinds coke into small particles ranging from 0 to 0.08 mm. This study illustrates methods of identifying defects in gear drive of the ball mill which is single line critical with respect to production.
The Ball Mill schematic (Make: McNally Bharat, Model: 37TPH) is shown in Figure 1 installed at a GAP. Sensor nodes enabling fault diagnosis for this Machine are mounted at positions marked in red dots. These sensor nodes communicate to a cloud dashboard and provided comprehensive vibration signal analysis and fault diagnosis.
3.1 GAP Mill Construction
The ball mill contains a hollow cylinder (filled with stainless steel or rubber balls) shell that rotates about its axis for grinding the material to fine form. Ball Mill uses a motor and gearbox to reduce the speed and stable operation to have high reliability and high efficiency over a long period in harsh environments. The final torque from Main Gearbox is transferred to a pinion and girth gear drive through a cardan shaft. Hence running this equipment under a fault condition is highly dangerous and could damage any drive component.
3.2 Sequence of Events
Figure 3 shows the sequence of events from fault diagnosis to corrective action.
Figure 3: Sequence of Events
The plots presented in this section validate the presence of the fault as predicted by the system.
- Figure 4 and Figure 5 show an increase in velocity & total acceleration trends.
- Figure 6 indicates the presence of Pinion/Girth gear defect frequencies & its harmonics predominating in the spectrum.
- Figure 7 shows an improvement in Total Acceleration after correction of root clearance of girth & pinion gear.
4.1 Advanced Fault Diagnosis in IDAP Platform
Shockwave and total acceleration are critical parameters which can be used to identify defects between pinion and girth gears like misalignment, improper meshing and bearing defects. Machine learning algorithms using these parameters were able to accurately predict gear defects well in advance in the GAP ball mill avoiding an estimated 48 hours of unplanned production downtime.