The art of measuring low-frequency vibrations

Rotor blades in wind turbines are growing longer – but also slower. Multimegawatt wind turbines turn even more slowly. This means that reproducible low-frequency vibration monitoring will gain in importance not only for the main rotor but also for the slow-operating gearbox components and roller bearings. Reliably measuring low frequencies, however, can be rather tricky. There are some notable aspects, which also pose new challenges to sensor and measurement hardware manufacturers. The following presents how PRUFTECHNIK Condition Monitoring GmbH is testing new products in its own test facilities as well as in wind turbine field trials.

Figure 1: Calculation of excitation frequencies (here, a section of the gearbox analysis)
Figure 1: Calculation of excitation frequencies (here, a section of the gearbox analysis)

 

1 Introduction

In terms of system size, wind energy is one of the industries experiencing the most „turbulent“ growth. In the past, system manufacturers only had very limited possibilities for testing wind turbines, which in turn had a somewhat negative effect on system availability. Condition Monitoring Systems (CMS) offer possibilities for collecting more information on operating and functional behavior.This works quite well in the fast and semi-fast-running drive train ranges and is the basis for vibration-based condition monitoring. There are difficulties in the slow-rotation range, though, and with low-frequency vibrations. It has happened that vibration measurement results could not be compared with one another, or that the wind turbines had highly perceptible vibrations and visible movements that could not be proven metrologically. Analyses and research findings showed that there were limitations to sensor systems, measurement hardware and the evaluation processes. New measurement methods, such as „resampling“, have meanwhile become the state-of-the-art – and not only for the condition monitoring of variable-speed wind turbines. Experience in the balancing of rotor blades in wind turbines under operating conditions shows that it makes sense to use low-frequency vibrations as an additional source of information for continuous condition monitoring. Within the scope of standardization work for VDI 3834 and ISO 10816-21, low-frequency vibrations were defined as an additional criterion for assessing the vibration states of slowly rotating component groups.

Figure 2: 30-s time signal of the vibration velocities at 6 rpm and 7 rpm and the associated frequency spectra up to 1 Hz plotted on a logarithmic amplitude scale.
Figure 2: 30-s time signal of the vibration velocities at 6 rpm and 7 rpm and the associated frequency spectra up to 1 Hz plotted on a logarithmic amplitude scale.

Figure 3: Different vibration and movement
Figure 3: Different vibration and movement
manifestations in a wind turbine
manifestations in a wind turbine

2 What are low-frequency vibrations?

According to VDI 3834 and ISO 10816-21, low-frequency vibrations are defined as vibrations between 0.1 Hz to 10 Hz. These vibrations are analyzed for their velocities and acceleration rates, and the least favorable values count. Just which excitation frequencies are excited by which component groups can be best determined on the basis of the kinematics and frequency tables in the CMS software. Figure 1 shows a section of the excitation frequency calculations for the first stage on a planetary gearbox. First, of course, the rotational frequency of the main rotor – with its numerous harmonics – must be considered. Then, the main rotor with its three rotor blades generates the rotor blade pass frequency (3p) with harmonic multiples. Rotational frequency excitations also originate at the main bearing, along with other particular excitations that arise from the roller bearing used (outer race, inner race, rolling elements and cage). At the gearbox input, most of the rotational frequencies of the first gear stage (LSS) and the roller bearing frequencies usually lie in the frequency range between 0.1 Hz to 10 Hz. In planetary gear sets, the number of planets and the associated rollover frequencies must also be taken into account. The extent to which the excitations from other gear stages then lie in the frequency range of 0.1 Hz to 10 Hz depends on the drive train design, the gearbox design and the rotational speeds in operation. In addition to these rotational-speed-dependent frequencies, the frequency range between 0.1 Hz to 10 Hz is more or less affected by the natural vibrations of the rotor blades, tower, nacelle and/or drive train. The frequencies and amplitude levels depend on the wind turbine design, and the monitoring center providing diagnosis services should have these data available. Advanced CMS, such as VIBGUARD XP, are capable of working as an FFT analyzer and determining these natural frequencies from temporary measurements, or even monitoring them over the long term to identify amplitude changes. For example, Figure 2 shows time signals of the vibration velocity and low-frequency vibration velocity spectra,measured at 6 rpm and 7 rpm directly in themhub of a 2 MW wind turbine. The comparisonof the frequency spectra shows how, at 6 rpm, resonance can develop between the rotor blade pass frequency and the tower‘s natural frequency. At 7 rpm, the 3p frequency lies higher and the tower‘s natural frequency at 0.33 Hz is not as strongly excited.

Figure 4: The generator speed and DC acceleration of a rotor blade in the circumferential direction while trundling and during subsequent acceleration
Figure 4: The generator speed and DC acceleration of a rotor blade in the circumferential direction while trundling and during subsequent acceleration
Figure 5: Influencing the DC acceleration in the wind direction with specific pitches
Figure 5: Influencing the DC acceleration in the wind direction with specific pitches

3 Low-frequency vibrations and movements around a rotating main rotor

 

Wind turbines are demanding and complex vibration and motion systems. Of
foremost interest are the mutually interactive motion characteristics of the „rotor and rotor blade“ drive machinery and how they are affected by gravity.
If gravity accelerometers are installed on the main rotor of a wind turbine, the rotor movements – e.g. in the centrifugal force direction – can be pinpointed as a function of direction. Figure 4 presents an example of a wind turbine start-up procedure. Shown are the generator rotational speed and a gravity acceleration in the centrifugal force direction, near rotor blade 3. When the accelerometer mounted on the rotor is at the bottom, then 1 g (9.81m/s²) is measured; when it is at the top, then -1g (-9.81 m/s²) is measured. If additional acceleration rates and tilting movements occur with respect to gravity, vectoral influences take effect. Using these measurements, it can be graphically tracked just how the rotor with its rotor blades accelerates, rotates or runs out. In the context of condition monitoring, such sensors can be used to detect whether it is always the same rotor blade, for example, that comes to a halt at the bottom due to an unbalanced mass. In addition to determining the rotational speeds and the acceleration, accelerometers such as these make it possible to evaluate additional position-dependent movements, angles of rotation of the individual rotor blades and the effective unbalanced mass. In fact, it is also possible to make in situ quality and acceptance checks of the rotor blade sets right in the wind turbine. If the gravity accelerometers are installed in the direction of the wind, the same sensors offer the possibility of measuring and tracking low-frequency movements and vibrations in the wind direction. Here, of course, the direction and positional dependencies in relation to gravity must be taken into account. Figure 5 illustrates how specific changes in the pitch angle influence the DC acceleration. This gives rise to possibilities for process control and for regulating the wind turbine.

Figure 6: Acceleration rates for rotor blade 3 in the
Figure 6: Acceleration rates for rotor blade 3 in the
centrifugal force direction (above) and in the wind
centrifugal force direction (above) and in the wind

In conjunction with condition monitoring, what is of interest is how the elastic rotor blades deform as a function of the wind speed and how consistently the three rotor blades move relative to one another. Using data collected over a period of one and a half day on one rotor blade, Figure 6 depicts these types of accelerations in the centrifugal force direction and in the wind direction, and also shows a plot of the wind speed. As wind speeds become stronger, the  acceleration changes based on the wind direction; this can be used for condition monitoring of the rotor blades. The question now arises as to how such varying gravitational acceleration rates can be explained. Answers are provided by operational deflection shape (ODS) analyses, where identically constructed sensors are installed in the pod, hub, rotor blades and drive train,as well as in the tower, and synchronized measurements are made at the same time and in the same phase. With regard to condition monitoring, such innovative measurement possibilities whet the appetite to obtain more information on the vibrational behavior at the upper and lower deflection points of the rotor blades. Thus, for instance, wear and tear in the pitch bearing mounts, electrical or hydraulic pitch errors, or individual rotor blade damage can all lead to runtime differences between the 3 rotor blades. These can be monitored using cameras or laser technology. Or, directionally-dependent additional vibrations and additional movements can occur. Gravity accelerometers make taking these measurements a lot easier.

 

Figure 7: Achieved rotor blade balance grades
Figure 7: Achieved rotor blade balance grades

4 Low-frequency vibration analyses in the stationary tower/pod system


In drive train condition monitoring, it has become practice to monitor the radial and axial pod vibrations in the frequency range of 0.1 Hz to 10 Hz on a continual basis.
If peaks become apparent in the statistical distribution, the causes should now be analyzed and eliminated if possible. How is this done?

 

Figure 7: Achieved rotor blade balance grades
can be monitored based on VDI 3834

Based on frequency and/or order analyses, diagnostic specialists first evaluate which component predominates when and how. Then, they try to identify natural frequency excitations by measuring runout amplitude spectra or using operational deflection shape analyses. If rotational frequency excitations dominate in the radial measurement direction, single-plane operational balancing will often correct the problem. If the tower‘s natural frequency is very prominent, however, finely-tuned balancing will be necessary to take as much energy out of the tower as possible so that the natural vibration in question is no longer excited. The necessary balance grade is achieved when the green range stipulated by VDI 3834 is reached
during the balancing process.
Just how precise are such quantitative vibration velocity analyses?The devil is in the details, particularly as the integration of very low-frequency vibration signals is no easy task. In the following section, the results of „the art of measuring low-frequency vibrations“ will be presented.

Figure 8: Characteristic lines of the hybrid triaxial accelerometer VIB 6.216 (top: characteristic line for the X and Y axes, bottom: haracteristic line for the Z axis)
Figure 8: Characteristic lines of the hybrid triaxial accelerometer VIB 6.216 (top: characteristic line for the X and Y axes, bottom: haracteristic line for the Z axis)

 

5 Sensors, measuremen technology and test options

Using hybrid accelerometers

Specifically for the wind energy sector, PRUFTECHNIK has developed a hybrid
accelerometer that can measure from 0 Hz (type VIB 6.216). The characteristic lines of the new Triax accelerometer and a picture of the sensor are shown in Figure 8. The sensor has no damping along the X and Y axes and its sensitivity is 500 mV/g. Users of accelerometers should note that only special accelerometers have such ideal characteristic lines. In the past, the most commonly used piezoelectric accelerometers could only measure from 2 Hz. This dropped to 0.5 Hz and, today, devices that can measure from 0.1 Hz are already available at affordable prices. Users should first have sensor manufacturers provide them with the characteristic lines of the sensors in order to see the magnitude of the damping. For instance, at first glance, deviations of 3 dB or 9 dB might not seem like much numerically, but when viewed in relation to pod vibrations of 100 mm/s, they would lead to erroneous measurements of 70.8 mm/s (3 dB) or even 35.5 mm/s (9 dB). This results in the rotor blade and pod vibration readings being too low on the measuring device by orders of magnitude. Roller bearing condition analyses are, however, different. Here, piezoelectric accelerometers are unparalleled; they make it possible to measure acceleration shocks of up to 100 g and can even handle the high-frequency resonance frequencies of rolling elements.
With the new hybrid Triax acceleration sensors, PRUFTECHNIK has combined the proven TANDEM-PIEZO technology with MEMS technology. This equipment uses a single sensor to perform not only machine diagnoses along the x, y and z axes, but also additional roller bearing diagnostics (along the z axis) and the previously described directionally-dependent acceleration rate diagnoses (along the x and y axes).

The measurement chain determines the quality of the results

Sensors alone are not enough to precisely measure low frequencies. DIN ISO 2159 and other standards specify that the entire measurement chain should operate with an inaccuracy of no more than 2 dB in specific frequency ranges. A measurement chain consists of the vibration sensor, the measurement system and the analysis software. It starts with the AD conversion and continues through to measurement processing and integration before finally providing vibration velocities. Hardware and software manufacturers require suitable testing equipment in order to adapt the entire measurement chain, right up to the software, to specific applications and to control it throughout.


Necessary quality controls

The quality of low-frequency vibration measurements can be examined internally and externally, and can even be confirmed by the PTB (Physikalisch-Technische Bundesanstalt, Germany‘s national metrology institute). In addition to „conventional“ measurement and test facilities, PRUFTECHNIK has of its own 3-axis motion simulator (Fig. 9), which makes it possible to rotate, position, wobble, tip, pivot and pitch items at freely-selectable rotational speeds, frequencies and signal waveforms. We will now report on verification measurements taken with VIBGUARD XP in the motion simulator, where positioning procedures, rotational movements and pitch movements can be simulated – much like in a wind turbine. In order to obtain an idea of which signal deviations can be expected in the low-frequency range, other accelerometers were also tested at the same time.

Figure 10: Positioning procedures in the motion simulator (left: acceleration rates in the gravitational vector; right: RMS and 0P vibration velocities at 0.1 Hz – 10 Hz)
Figure 10: Positioning procedures in the motion simulator (left: acceleration rates in the gravitational vector; right: RMS and 0P vibration velocities at 0.1 Hz – 10 Hz)

Figure 10, on the left, shows simple positioning procedures in the form of gravity acceleration, and on the right, it shows them as RMS and 0-p vibration velocity in the frequency range of 0.1 Hz to 10 Hz. The respective sensor positions are read from the directionally-dependent acceleration operations, and the stresses that occur when the motion simulator is accelerated or slowed are determined using the vibration velocity. By comparing the amplitudes, it is even possible to detect the influences of residual imbalance.
Vibration velocity analyses in the motion simulator are handled similarly to those in a wind turbine.

Figure 11: Time signals and frequency spectra at 6 rpm and 9 rpm in the 3D motion simulator
Figure 11: Time signals and frequency spectra at 6 rpm and 9 rpm in the 3D motion simulator

Figure 11 shows the vibration velocity time signals at 6 rpm and 9 rpm and the frequency spectra computed in VIBGUARD XP in a logarithmic diagram.


 

Figure 12: RMS and 0P vibration velocity values at 0.1 Hz – 10 Hz for three different accelerometers (simultaneously measured at different rotational speeds)
Figure 12: RMS and 0P vibration velocity values at 0.1 Hz – 10 Hz for three different accelerometers (simultaneously measured at different rotational speeds)

Figure 12 shows a comparison of continually recorded RMS and 0-p vibrationvelocities in the frequency range of 0.1 Hz to 10 Hz while the system was inmotion at different rotational speeds in the motion simulator. The diagrams reveal that the signal levels of the right sensor deviate significantly, giving us reason to doubt the linearity here.

 

6 Outlook

The experience and examples described here demonstrate that manufacturers of measuring and testing equipment have the potential and the possibilities to make new technologies available to the wind energy sector in the form of innovative and hybrid solutions. Even the slow-operating „rotor with rotor blades“ drive machinery can now be measured and monitored using the new hybrid triaxial accelerometers described above. Manufacturers and operators of wind turbines have the option of extending the conventional condition monitoring of the drive train to the main rotor and rotor blades. In future, it should even be possible to use acceleration signals in single-pitch systems for low-vibration pitch actuation.

 

7 Literature

[1] www.telediagnose.com , Issue 5 and Issue 12
[2] Becker E., Lösl: “Unwucht birgt Gefahren für Komponenten- Das Auswuchten von Rotorblättern kann Schwingungen mindern” (Imbalance poses hazards to components; rotor blade balancing can reduce vibrations); Z. Erneuerbare Energien (Renewable energies publication), 08-2009
[3] VDI 3834 Sheet 1 “Messung und Beurteilung der mechanischen Schwingungen von Windenergieanlagen und deren Komponenten – Windenergieanlagen mit Getrieben“ (Measuring and assessing the mechanical vibrations of wind turbines and their components; wind turbines with gearboxes)
[4] ISO DIS 10816-21:2013 (E): “Mechanical
vibration evaluation of machine vibration by measurements on non-rotating
parts; Part 21: Horizontal axis wind turbines with gearbox”
[5] Germanischer Lloyd Industrial Services GmbH, Wind energy business sector, 2013: “Richtlinie für die Zertifizierung von Condition Monitoring Systemen für Windenergieanlagen” (Guidelines for the certification of condition monitoring
systems for wind turbines) Windenergieanlagen“

Do you like this application?

Enquire now!

 

 

Contact us