I have question of why is it convenient to classify faults in gearboxes as being either localised or distributed?

I see some reference in the attached to distributed, but not local; so I'm still looked for the convenience angle answered. It has been shown that using these ways of vibration signal analysis there are possibilities to detect signal faults and distributed faults in gearboxes. A signal fault is caused by a tooth crack/fracture and breakage, a spall in a gearing or in an inner or outer race of a bearing, a spall on a rolling element of a bearing; distributed faults are caused by uneven wear (pitting, scuffing, abrasion, erosion).

Gearbox_Diagnostics_Fault_Detection.doc (317 Kb, 45 downloads)

My layman's guess is so that it's easy to study and treat them. Corrosion can be cassified as localised or uniform/distributed so that it's easy to monitor and calculate.

R u a gearbox specialist?

P/s Why did you start a new thread on this same topic? Try not to break the thread for the same topic for easy reference in the future for those who need it.

quote:

R u a gearbox specialist?

No; I'm not a mechanical engineer; simply faced with such q's at the moment as part of some study I'm doing

Which form of damage has worst effects, localised or distributed?

Here is probably a relevant study:
http://www.rpi.edu/~lic3/James_Li_RPI/PDF/MaNewApprGear...signal%20modulation'

Here is some work drafted on this......

- The mid frequency range is dominated by features associated with gearboxes. Subtlety of techniques is required to detect and diagnose a range of common gear faults, which can be segregated as localised or distributed. Local meaning specific to a gear whereas distributed is beyond a singular gear

- Gears are made to a very high degree of accuracy, to ensure that the motion transmitted is purely rotational and yet when put into service the resulting vibration spectrum exhibits harmonics of tooth mesh which can only be coming from distortion – demonstrating that the motion transmitted is not pure rotation. Harmonics come from distortion

- Localised amplitude modulation. Early-stage fatigue can be detected by relying upon the fact that the tooth suffering the fault will be slightly weaker than the remaining healthy teeth on the gear and it will hence deflect more than the other teeth during application of load in the mesh. As a result of the weakness of the tooth, which is suffering the internal fatigue, damage the excessive deflection that it exhibits will cause localised regions of non-uniformity within the otherwise uniform tooth-meshing signal. The amplitude of the tooth meshing vibration will increase when the faulty tooth is in the mesh (this is called amplitude modulation) and since it only occurs in localized regions we refer to it more specifically as localised amplitude modulation

- Just like the occurrence of distortion is synonymous with the presence of harmonics, modulation (of any sort) is synonymous with the presence of sidebands

- Sidebands appear in a spectrum as a family of equal-spaced peaks. We measure the sideband spacing. This tells you the frequency at which the modulation is occurring (if there is only a single fault on a gear, this will be the gear's' rotational frequency

- This is extremely important and very powerful because it enables you to identify which gear (or pair of gears) is causing the modulation – and hence is suffering the fault

- If the sidebands are low in level and extend broadly across the frequency range then they are indicative of local modulation

- All the sidebands will be at exactly the same spacing and this should be measured accurately because it will relate precisely to the exact modulation frequency



Other common faults in gears also give rise to modulation of the tooth mesh signal but in ways, which are, different to the effects of localised bending fatigue. Such faults include eccentricity, being bored off centre, and a bent gear shaft. These are then referred to as distributed

- These types of faults result in distributed amplitude modulation

- This is another form of modulation it will give rise to sidebands in the spectrum



- In this case the shape of the sideband pattern is that the sidebands are now tall and clustered around the tooth mesh fundamental and its harmonics (when talking about modulation are often referred to as the carrier frequency and its harmonics). The shape of the sideband pattern is now very different to that from localised modulation and this ability to infer the nature of the modulation from the shape of the resulting sideband pattern is a very powerful diagnostic tool

- The shape of the sideband family tells the nature of the modulation and allows us to deduce local or distributed types of faults

- The symptoms described here relate to incipient faults. Gears sideband pattern is quickly lost as faults progress from incipient to more serious state

- Reference [] Gearbox Diagnostics Fault Detection By Walter Bartelmus discusses the concept of signal faults and distributed faults on gearboxes as follows: - "The paper deals with the method of gearbox diagnostics fault detection, and shows that using: design, production technology, operational, change of condition (DPTOCC) factors analysis leads to inference diagnostic information

- The paper shows that for gearbox fault detection, many different ways of signal analysis should be done. In the paper it is suggested that for fault detection: time trace, spectrum, cepstrum and time-frequency spectrogram examination has to be used. It has been shown that using these ways of vibration signal analysis there are possibilities to detect signal faults and distributed faults in gearboxes. A signal fault is caused by a tooth crack/fracture and breakage, a spall in a gearing or in an inner or outer race of a bearing, a spall on a rolling element of a bearing; distributed faults are caused by uneven wear (pitting, scuffing, abrasion, erosion)

- Computer simulation enables one to infer that the cepstrum not only detects single gearing faults, but also distributed faults. It is pointed that for explicit detection of a tooth fracture and breakage there is a need to use a cepstrum and a time-frequency spectrogram

- Condition change factors include condition of bearings and gearbox gearing and shafts. For gearing we have signal faults (cracking/fracture, breakage) and distributed faults, pitting, scuffing, erosion. For bearings we have signal faults (spalling, pitting, erosion) occurring on bearing's races and rolling elements. The bearing elements undergo also abrasive wear causing bearing elements dimensions change

- Computer simulation investigations on influence of distributed faults to cepstrum show that cepstrum is also a measure of distributed faults"

- Reference [] A Model-Based Gear Diagnostic Technique Wenyi Wang and Albert K. Wong talks about localised gear faults with: - "Currently, most methods developed for vibration-based condition monitoring of power trains are based on the processing of the vibration signals through various means of filtering and conditioning. Whilst this process has evolved into a well established procedure, its application still requires a high level of expertise in the selection of filter bands, spectral lines and other components to be removed, etc. This has been an obvious hindrance for automation of fault detection. To overcome this, a model-based method, utilizing autoregressive (AR) modelling, is proposed

- An AR model is established for the synchronous signal average under the healthy conditions of the monitored transmission and is used as a linear prediction filter. The future signal averages, under healthy or faulty conditions, from the same transmission gearbox are then passed through the AR model filter. Subtracting the filtered signal from its original version can generate an AR model residual signal. This residual signal will only reflect the prediction error of the AR model and be randomly distributed if the monitored gearbox remains in a healthy condition. However, when a localised gear fault, such as a tooth crack, is developed within the gearbox, the fault-affected area will not be well predicted by the AR model that was generated under healthy conditions. Consequently, the AR residual signal will reflect the sudden changes caused by the local fault


- The study on incipient failure detection of gearboxes started over two decades ago and achieved some major breakthroughs with the advent of the high frequency resonance analysis technique for rolling bearings and the synchronous signal average technique (SSAT) for gears. For gear fault diagnosis, the most commonly used techniques are derived from, or based on, the SSAT. The SSAT and its three derivatives, i.e., the residual signal analysis; the narrow-band demodulation technique; and the time-frequency analysis technique, will be briefly discussed

Current Gear Diagnostic Techniques

- The SSAT is widely regarded as a powerful tool in the detection and diagnosis of gear faults. It forms the basis of current gear-fault diagnostics. The residual signal analysis on the SA is the most popular technique for gear fault diagnosis. The narrowband demodulation technique is a derivative approach of the SSAT. It concentrates on the extraction of diagnostic information from the low-energy high-order modulation sidebands associated with the largest meshing harmonics in the spectrum of a SA. The time-frequency analysis approach separates the gear meshing components from the localised phenomenon associated with gear faults. These three methods are: -

1. Synchronous Signal Averaging Technique
2. Narrow-Band Demodulation Technique
3. Time-Frequency Analysis Technique