Various types of fans are used in cement production. Despite the rather simple technology of these machines, carrying out only corrective maintenance is ineffective and can result in long and costly shut down of production lines.

Consequently, maintaining the stable operational conditions of fans by implementing preventive maintenance is of utmost importance for the overall economy of cement plants. In this article, the main aspects of preventive maintenance as applied to centrifugal fans will be addressed and special attention will be paid to conditional maintenance.

Figure 1 illustrates the definition relative to maintenance.

Periodic maintenance A typical example of periodic maintenance is the lubrication of the rotor bearings at regular intervals. When lubricated by grease, the amount of grease for each relubrication is given by the following formula:

G = 0.005 Y D Y B

where: G=grease quantity (grams)

D=bearing outside diam (mm)

B=total bearing width (mm)

The intervals obtained from the resulting diagram should be halved for every 15 degrees C increase of bearing temperature above 70 degrees C. All other instructions relative to the periodic maintenance are given in the maintenance manual delivered with the machine. For example: * B the control of the fixing bolts tightening; * B the impeller cleaning; * B the wear-liners repair; and * B the fan rebalancing after repair.

If a recommended time period is given by the vendor, it has nevertheless to be adjusted from one plant to another, as operating conditions vary.

Conditional maintenance Conditional maintenance depends on a specific type of event (self diagnosis, information from a sensor, measurement of a wear rate, etc.), showing the condition of degradation of the goods.

The clear advantage of this method is that intervention must be carried out only if some parameters have a significant evolution. Obviously this involves the operating parameters being periodically measured in order to anticipate, diagnose, and finally plan a repair before the failure (see Figure 2). Bearing vibrations should be periodically checked in order to predict either the bearing degradation, or a change in balancing quality as a result of wear, build-up, etc.

Detection of the defect When equipment is commissioned, the basic parameters are recorded in order to establish a reference to be used for later comparisons.

For machines such as process fans, the data to take into account are the vibration and temperature of all bearings and the operating conditions such as:

* the opening of inlet vanes (dampers); * the flue gas temperature; * the dust load; * the flow-pressure operating point; and * the motor absorbed power.

Diagnosis As soon as a deviation is detected, the maintenance engineer has to determine, if possible, the source and severity of the suspected defect.

Analysis of trend Depending on the deviations of readings and diagnostics, the trend should be extrapolated to estimate failure time and the repair made at a convenient time. The interval between readings is then reduced.

Analysis of vibration Monitoring vibration to control rotating machines has been carried out for many years. Presently, techniques are being upgraded making vibration analysis the most useful implement for modern conditional maintenance. Mechanical vibrations are oscillating movements, either periodic, non-periodic (transitory or random), or both.

All periodic signals can be described in function of time as follows:

Consult Printed Publication for Formula

The spectrum is the representation of lines of amplitudes versus frequency (Figure 3).

It is worth pointing out that the notion of phase disappears when directly using spectra of amplitude. In other words, two different analyzed signals, composed of the same two harmonics with different phases, have the same resulting spectrum.

In Figure 4, some characteristic spectra of typical vibration signals are shown as examples. The purpose of the spectral analysis is to identify some of the lines (or peaks) composing a spectrum as possible defects, which can appear in the shape of three categories: peaks at multiples or low-multiples of rotating speed; peaks independent of rotating speed; and spectral density coming from random components of vibration. Spectral analysis can be used to identify the cause of most common defects in fans. These defects are detected by typical spectrums.

Unbalance Unbalance is one of the most common sources of vibration. It can be caused by build-up, temperature excursion, or wear on the impeller. The spectral analysis shows a peak meanly at the speed of rotation (harmonic 1).

It can be caused by axial or angular misalignment, or a defect on coupling: teeth, springs, covers, greasing, geometry. The spectral analysis is characterized by a significant number of harmonics and often the harmonic 2.

Fixing defects When loose fixing, either bearing bolts or anchoring bolts, the spectral analysis is a succession of harmonics due to a time signal, which is almost rectangular (Figure 4).

V-belt defects The degradation of one or several belts induces variations of tension between two pulleys and, therefore, vibrations. The spectral analysis shows peaks at frequencies that are harmonics (1 to 3, sometimes more) of the belt rotating speed.

Rotating stall One typical fan aerodynamic phenomena that can be analyzed through the spectral analysis is the rotating stall, which occurs at low speed and can be represented by regions of flow blockage (or stall cells) inside the impeller (Figure 5).

"Consult Printed Publication for Formula"

N[subscript]1=rotating speed of pulley 1 N[subscript]2=rotating speed of pulley 2 d[subscript]1=diameter of pulley 1 d[subscript]2=diameter of pulley 2 L[subscript]belt=total length of belt

What is typical is the fact that the spectral analysis shows peaks at frequencies equal to 21/43, 41/43, 51/43 the frequency of rotation.

Bearing condition analysis Bearings are among the components of a fan that are subjected to the most severe wear. In this section, typical vibrations from defective bearings is described, and the most important analysis methods presented: Crest Factor and Envelope Detection.

Despite a minimum calculated lifetime (e.g. 80,000 hours), bearings can be damaged earlier because of transport or storage (vibration, water condensation); mounting (shocks, arc welding, residual clearance, shaft dimension, alignment); or service conditions (overload, vibration, lubricating defect, overheating, penetration of dust or water, quality of bearing housing).

Vibration from a defective bearing Figure 6 represents a local defect on an outer ring. When a roller or a ball passes over a defect, it creates a shock on the ring and consequently a repetitive impulsion, with an amplitude proportional to the shock intensity (speed, clearance, load and local defect).

"Crest factor" method The principle illustrated on Figure 7 (page 59) is to compare the peak level (sensitive to shocks) and effective RMS level (average energy) of acceleration time signal, by follow-up of the ratio Peak/RMS ratio (equal to 1,414 for an ideal sinusoidal signal).

Light defects on bearing make the peak level increase without significantly affecting the RMS level. Thus, the crest factor increases with defect development.

Experience shows that the crest factor begins at about 3, then increases, and tends to decrease for advanced defects. This method can be applied with reliability, only with a periodic follow-up, and mainly at an early stage.

Envelope detection Most modern analyzers or data collectors offer envelope detection, which can be useful for diagnosis.

The purpose is to extract the periodic signals (shocks) at very low amplitude within high frequencies. This is the only way to avoid the background noise and random shocks. A clear picture of the damaged bearing is obtained with the envelop spectrum.

Conclusion The numerous advantages and possibilities of preventive maintenance, especially conditional maintenance, can be the source of major cost reductions and may implemented in all sections of a cement plant.

Reducing the costs of emergency repairs and spares storage; increasing production through better availability of machines; scheduling of repairs in accordance with production imperatives; improving safety through periodic check-ups of machines; and increasing credibility of the maintenance department are all results of this type of action

Conditional maintenance involves the periodic measurement of a specific parameter-namely vibrations through the use of spectral analysis. The main types of typical faults were: unbalancing, misalignment, fixing or coupling defects, aerodynamic defects, and bearings. The bearing temperature is another important parameter to monitor. Of course, all of these measurements have to be correlated with operating conditions for interpretation.

Herve Bousson is maintenance chief engineer and Alain Godichon is technical director, both for ABB Solyvent-Ventec.