Rotary kilns are one of the most failure-prone parts of a cement plant. Not only do they operate under a high degree of thermal and mechanical stress, they also require prompt corrective measures to mitigate safety hazards, and avoid lengthy, unplanned downtimes.
However, the difficulty of preventing common failure modes in time is linked partly to the complexity of interaction between the kiln and mechanical parts like the drive system and trunnion components. It is further exacerbated by the size of the system, adverse operating conditions, and lack of accessibility to internal components without long shutdowns. Considering that a kiln must run continuously for a year before a shutdown, manual inspection of internal components is unviable.
This is where predictive maintenance can prove invaluable to businesses in the cement industry. By exploiting advanced analytics techniques, IoT sensors, and real-time data streams, predictive maintenance can have a consequential impact on your plant’s uptime. In this article, take a look at the top rotary kiln maintenance issues, and how predictive analytics can help you overcome them.
Optimizing kiln maintenance processes with predictive analytics
Preventing refractory failure in a kiln with thermography and computer vision
Refractory failures are usually associated with overheating of the kiln. This may occur due to excessively rapid heating of the kiln, erratic feeding and advancement, deviation in the refractory shape, or even due to lack of consistency in the chemical makeup of the feed. Other reasons can also cause refractory failure – for instance, colling the kiln too fast, very frequent shutdowns, or poorly directed flame patterns.
There are multiple predictive techniques for preventing refractory failures. One way is to use pyrometric cameras inside the kiln, and send heat signatures captured at a regular interval to a predictive maintenance platform. These signatures are then analyzed by machine learning algorithms to detect spontaneous spikes in temperatures, and alert operators in time.
Another way is to build a 3D model of the kiln using LIDAR sensors, and map inputs from infrared sensors installed outside the kiln onto the model. This can help determine the precise location of rise in temperature, and diagnose issues like locational loss of thickness inside the kiln. Moreover, AI models can be trained to determine the thickness of the wall based on heat signatures, and this can help operators react to reducing brick thickness in time.
Detecting excessive kiln shell ovality to prevent lining damage by monitoring tyre migration
While the thin-walled design of a kiln is crucial to minimize its weight, it also results in ovality. After startup, when the tyre and the kiln shell reaches a thermodynamic equilibrium, the top clearance reduces to accommodate a safety margin, which shields events of brief overheating. However, if the top clearance is too much even after the system has been heated after startup, the kiln will attain a high degree of cross-sectional ovality. During rotation, this causes excessive flexing of the brick lining, which can contribute to shell damage. Moreover, increased cyclic fatigue will also result in shell cracks near the tyre.
What causes excessive kiln shell ovality?
High shell ovality can be caused by numerous factors. These include wear of tire support pads, a dogleg condition spanning multiple piers, suboptimally sized tires resulting in inadequate support to the cross section, or excessive tire elevation.
While it is possible to detect ovality with ovality sensors, (they are attached to the shell tire), it is not possible to infer the root cause of increased ovality with their output alone.
For this reason, it is crucial to use ML models which can analyze numerous input parameters like tyre migration, thermal conditions, and changes in ovality, to inform operators of the exact cause of the problem. This enables prompt corrective action, and can prevent costly conditions like permanent shell deformity.
Preventing and counteracting shell cranks with ML-based parametric analysis
Crank is a condition of eccentricity between the axis of rotation, and a part of the kiln shell’s centreline axis. It manifests as curving of a section of the shell, and this causes mechanical stress through additional cyclic load on rollers, tyre, and shaft.
What causes cranks in the kiln?
Kiln cranks can be caused by thermal or mechanical factors. For example, variable brick thickness or uneven buildup of coating can cause specific areas of the kiln to become warmer, resulting in an outward longitudinal curve. These are called thermal cranks, and are usually temporary. On the other hand, incorrect assembly or welding errors can result in a mechanical crank.
Mechanical cranks tend to be permanent, and require correction before operation – especially because even minor cranks will cause significant change in cyclic load and ultimately result in kiln failure. Kiln crank can be measured through cyclic deflection of the supporting shafts.
Mitigating thermal cranks
Thermal cranks can be corrected, given that the clinker production process is monitored closely for multiple parameters. When thermal signatures, tyre displacements, and bearing vibrations are simultaneously fed to pretrained ML models, it is possible to identify the root cause of the crank condition.
For example, thermal cranks can be counteracted with localized cooling (however using water for cooling will result in shell damage), whereas unfavorable mechanical conditions can be eliminated with parametric analysis.
Predictive kiln monitoring calls for advanced analytics
Kiln failure modes can be difficult to diagnose, especially because of the complex interactions between components that define the state of a rotary kiln at a given time. For example, girth gear connector looseness, kiln dynamics, and shell deformities, can all contribute to drive vibration.
This necessitates the use of advanced ML-based predictive systems, which can correlate multiple, real-time data streams simultaneously, and emit the root cause of anomalous behavior in advance.
Next steps
Being a central component in a cement plant, a failure of the rotary kiln can lead to long shutdowns, no matter what they are caused by. Cement plants typically face 9-10 kiln failures in a year. This, in itself, is an indicator of the complexity associated with anticipating and preventing kiln failures.
In other words, kiln predictive maintenance represents a significant opportunity for cement manufacturers. Advance your kiln maintenance workflows today with an industry-leading predictive maintenance platform. Contact UptimeAI now.