In the power industry, fuel costs make up a large portion of the operating expenses associated with the generation of electrical energy. In addition to the "per ton" cost of raw coal delivered to a power-generating site, fuel preparation costs must also be considered. Fuel preparation involves the crushing of large chunks of coal, followed by pulverization to achieve desired fineness for proper combustion. Since coal fineness is critical for proper combustion, close monitoring of the operating characteristics of individual pulverizers would be highly desirable.  Prior to development of this dynamic analysis system, operating assessment of pulverizers has been limited to personnel judgment based on control system variables such as coal and airflow rates, motor amps, temperature and pressure. Anomalies with the grinding elements within the mill are often only detectable by noise or vibration when the problem becomes severe. With the introduction of this dynamic analysis system, operating characteristics of individual pulverizers can be closely monitored, and operating & maintenance crews are provided with the ability to optimize pulverizer performance and improve combustion efficiency.

Another important factor in profitable operation of coal fired power plants is the prevention of forced outages.  Since large coal pulverizers can process over a thousand tons of coal in a 24 hour period, component erosion is an on going problem.  The synergistic effects of component wear lead to degradation of pulverizer performance, and often result in catastrophic component failure.  Visual inspections and measurements taken while the equipment is out-of-service are presently used to determine condition of pulverizer components.  With the introduction of this dynamic analysis system, component condition can be closely monitored and load settings of individual grinding components can be balanced.  Installation and adjustment problems can be identified at the time of equipment start-up, and catastrophic failures can be prevented by early problem detection.

Dynamic Analysis Method

The dynamic analysis system provides a tool and methodology for analysis of pulverizer components and pulverizer operation.  Through the use of this system, operational problems can be identified, component wear can be predicted and catastrophic failure can be reduced or prevented.  The dynamic analysis system is comprised of displacement-monitoring transducers, signal-conditioning equipment, a computerized data acquisition system, and related software configured for installation on existing coal pulverizing equipment(Figure 1).  In this arrangement, displacement transducers are affixed to loading members of pulverizer grinding components to provide a continuous motion analysis record for the component.  Conditioned signals are collected in a data acquisition system, and proprietary software for the system provides the analysis capabilities to predict numerous component and operational problems associated with the equipment.

The data acquisition system is typically formatted to collect pulverizer loading amps and displacement signals for each pulverizer-loading component.  Proprietary software associated with the data acquisition system collects & archives the data, and allows test data to be viewed on the industrial hardened computer monitor.  Base line data, collected for a wide variety of different operating conditions (start-up, increasing load, steady state, decreasing load, & shut-down) can be saved for comparison with later operating conditions to predict wear rates and identify potential problems.

In a typical analysis scenario, a variety of different data plots are employed to determine operating conditions and identify problems.  Time Domain Analysis methods can be applied to optimize loading conditions in each mill, and Frequency Domain Analysis methods can be applied to determine the condition of individual components and to analyze operational problems.

Site Installation

The dynamic analysis system was first applied on four units at Ohio Edison's WH Sammis Generating Station, in 1993.  As main shaft failures had been an ongoing problem at this facility equipped with Raymond Bowl Mills, the dynamic analysis system was initially installed to address this problem, and has since been used to monitor grinding element wear and to insure component structural integrity.

Installing the displacement transmitters at the loading components is accomplished by first welding mounting studs to the loading components.  With the mounting studs installed, displacement transducers are fastened to the mounting studs and calibrated.

Installing the local data acquisition & analysis computer is easily accomplished as the equipment has been incorporated in a rolling storage unit for portability and convenient set-up.  After rolling the unit to the test location, the computer monitor & keyboard are removed from their storage locations, and set in their operating positions at the top of the storage unit.  The local data acquisition & analysis unit is then ready for connection to a 120V power supply.

Wiring of the transducers and current indicating transmitter is the final step in the installation.  Wiring of the transducers is accomplished with quick disconnects at the termination unit.  Wiring of the current indicating transmitter can be terminated at local instrumentation or at the pulverizer motor control center.

Following initial installation of the transducers and wiring of the pulverizer current indicating transmitter, installation of the dynamic analysis system will take approximately ten minutes.

DETAILS

Methods of Analysis

In a typical analysis scenario, a variety of different data plots are employed to determine operating conditions and identify problems.  Among the most informative data plots are the Relative Displacement, Loading Amps vs.  Displacement, Power Spectral Density, and Profile plots.  Each of these data plots has specific benefits for analysis and can be used to detect specific problems.

The Relative Displacement plot is used to determine the relative loading conditions of each grinding component (Figure 2).  By comparing the absolute displacement of one grinding component to the others, imbalances in loading can be identified, and adjustments can be made as required.  Since unbalanced loading can create cyclic stresses in the yoke and main shaft assemblies, balancing the loading conditions is an important first step in reducing component failures (such as fatigue failures in main shafts).  Coal bed depth is generally kept to less than ½" as bed depths much deeper than this usually result in reduced fineness and increased spillage.  In the Relative Displacement plot (Figure 2), the deflection of Roll 1 can be seen to be nearly twice that of the other two rolls.  This condition, identified by the system, was corrected by a “zero" adjustment on Roll #1's journal assembly.

The Loading Amps vs.  Displacement plot is used to compare average grinding component displacements taken at different capacities (Figure 3).  This allows comparison of average displacement between mills, as well as comparison of one mill's present average displacement with previous conditions.  Through this method of analysis, operations crews are able to analyze the grinding performance of one mill, with that of others, providing indications as to where the most cost effective efficiency improvements could be realized.  This analysis method also allows maintenance crews to compare present grinding operation with past operation and provides an indicator to the maintenance requirements of each individual pulverizer.  In the Loading Amps vs. Displacement plot shown (Figure 3), data taken from Mill 1B after a pulverizer rebuild is compared with data from other mills on the same unit.  Notice that the slope of the line for Mill 1B data (red line) is steeper than the slope of the line for data regressed from other mills on this unit (green line).  This plot illustrates graphically that Mill 1B's springs are stiffer than average.  The minimal effect of increased classification on pulverizer power requirements is also depicted on this plot.  Increasing the classifiers at a given load to obtain a finer product tends to shift the data point up, roughly along the same Amps vs. Defection slope line.

The Power Spectral Density plot (Fast Fourier Transform squared) is used to analyze the dynamics of individual pulverizer components under operating conditions (Figure 4).  In addition to allowing dynamic comparison of individual components within a mill, this plot also provides for the detailed analysis of operating anomalies in the mill.  By identifying the frequency of relatively high amplitude spikes on this plot (greater than .200), operating anomalies can be traced to individual components, allowing for timely replacement of worn components prior to catastrophic failure.  In addition to the obvious benefit of relating components to their respective problems, this method of analysis also provides an invaluable tool in predicting wear rates for specific components, estimating remaining life and prioritizing maintenance expenditures.  In the plot shown (Figure 4), the power spectral density of each of three rolls of a CE Raymond Mill are displayed.  Each spike on this plot is identified by it's driving frequency component.  Note here that the three individually filtered roll waveforms sense the large excitation of the bowl, which is turning at approximately 82 Revolutions per Minute.  The large amplitude spike at approximately 190 Revolutions per Minute is sensed only by Roll 3 and is hence an anomaly related to that roll.  The fact that rolls wear over time is illustrated by a frequency shift.  As a roll wears, its diameter gets smaller causing it to spin proportionally faster.  This speed can be manipulated and used to track the wear of the roll over time.  Large amplitude spikes are usually indicative of mechanical problems within the mill.  For instance, a large 1X spike would occur if a main shaft had a small crack.  For each revolution of the bowl, the crack on one side of the shaft would open when the bowl was aligned 180 degrees from an opposing roll's force.  When the bowl turned such that the crack was in line with the roll force, the crack (then in compression) would be masked.

The power spectral density of a mill with a grinding segment out of place is similar to a mill with a cracked main shaft.  Each time the bowl turns the protruding liner segment causes a blip in the waveform.  The FFT algorithm identifies this properly as 1X excitation, however one needs to investigate further to determine the forcing function.

Profile plots combine time and frequency domain information in a manner that is particularly useful for visual imaging of the grinding zone (Figure 5).  In these plots, data points are placed on a polar coordinate system with linear displacements displayed as a radial dimension extending out from the origin.  With the displacements magnified for greater resolution, these plots display a pattern associated with the general shape of the grinding zone.  A perfectly uniform bowl displayed as a Profile plot would appear as a perfectly uniform circle.  In the Profile plot shown (Figure 5) three distinct lobes are evident.  These three lobes represent increased roll displacements, which occur three times during each revolution of the mill.  By revolving the image in the Profile plot about its central axis, you can see what a fixed roll would sense during each revolution of the mill.  When used in conjunction with the Power Spectral Density plot (Figure 4) which depicts a spike at three times the bowl frequency (~ 250 RPM), the Profile plot (Figure 5) confirms this data and allows the engineer to more clearly visualize the movement of the roll.

Raymond Bowl Mills

The dynamic analysis system can be applied to Raymond Bowl mills with transducers installed at the spring frames of the journal assemblies.  Through the use of this system, spring integrity and individual loading rates at the journal assemblies can be determined.  Unbalanced loading conditions and resultant problems such as stress concentrations applied at the main shaft can be identified and corrected.  Since stress concentrations at the main shaft can lead to fatigue failure of these components, the dynamic analysis system provides a convenient method for preventing such failures.  Roll and bowl wear can also be monitored and wear rates can be determined while the mill is in service.  This allows component replacement schedules to be incorporated into preventive maintenance outages and effectively reduces the likelihood of component forced outages.

El Pulverizers

The dynamic analysis system can be applied to EL Pulverizers with transducers installed at the grinding ring loading assemblies.  Through the use of this system, unbalanced loading conditions can be identified and corrected, reducing or eliminating main shaft fatigue failures.  Ball and grinding ring wear can also be monitored and the dynamics of ball sets can be analyzed to detect orbiting and skidding problems.  By detecting & correcting these primary problems early, the secondary problems of abnormal ball wear, ball breakage, ring grooving and ring chipping can be prevented.  Since wear rates and component condition can be determined while the mill is in service, replacement schedules can be incorporated into preventive maintenance outages and component forced outages are effectively reduced.

MPS Pulverizers

The dynamic analysis system can be applied to MPS Pulverizers with transducers installed at the pressure frame assembly.  Through the use of this system, loading rates can be determined and unbalanced loading conditions can be identified and corrected.  By providing a reference indication for spring loading pressure at each roll wheel assembly, the system also provides for analysis of on-line loading pressure adjustments.  Dynamic problems such as roll wheel vibrations can be analyzed through the use of this system, and proper corrective measures can be taken prior to extensive component damage.  This on-line analysis capability provides a valuable tool for the reduction of forced outages, and along with the ability to track wear of roll wheels and grinding elements, allows component replacement schedules to be incorporated into preventive maintenance outages.

Cost Savings Anticipated