Elastohydrodynamically lubricated (EHL) contacts are fundamental for severalrotating machine elements. For example gears, rolling element bearingsand lubricated chain drives work due to the principle of EHL. All of these machineelements require maintenance and, as condition based maintenance hasincreased in industry, the need for monitoring techniques has also increased. Inorder to avoid incorrect condition indications, since the 60’s researchers haveimproved signal processing of existing monitoring tools and developed newtechniques as a complement to these existing tools. In the past two decadesacoustic emission has been identified as a new complementary tool for monitoringof rolling element bearings and investigated intensively by several researchgroups. However, most of the investigations were carried out at lowrotational speeds. Furthermore, most of the investigations used simple signalprocessing methods like activation counts (AC) or trend analysis of the rootmean square signal (RMS). One reason for using simple experimental conditionsand signal processing methods is the complexity of a rolling elementbearing itself. A rolling element bearing consists of several EHL contacts andeach contact has different operational conditions (film thickness, slide to rollratio, contact pressure, entrainment speed). The measured signal is the summationof all EHL contacts. This complexity is one reason why the high frequencyemissions of an EHL contact are still not fully understood. Therefore, an investigationof the acoustic emission of a single EHL contact was here carriedout within the framework of a PhD project.In this thesis simplified experiments were used to represent either a single EHLcontact or elements of an EHL contact. Both acoustic emissions of tensile testsand ball impacts on a solid plate were studied and analyzed with respect totheir significance for EHL contacts. For all investigations carried out in thisthesis an absolute calibration method developed by McLaskey and Glaser wasused. This calibration method was validated for boundary restricted systems,where a good agreement for zero frequencies was found, however, unsatisfying agreement was discovered for resonances of a boundary restricted system. Theinvestigation found elastic waves in boundary restricted systems consist of twofundamental types. Zero frequencies will be enhanced for cases were excitationsource and elastic wave are independent, while an interaction of sourceand elastic wave results in a pure resonance problem.Furthermore, the time dependency of acoustic emission signals was investigated.As mentioned previously most existing investigations are carried outat low and constant rotational speed. The dependency of acoustic emissionsignals and speed is reported in literature as well as difficulties with acousticemission measurements at elevated rotational speeds. By using ball impactswith different ball sizes and tensile tests with different displacement speeds thetime dependency was analyzed with respect to excitation time (contact time ofball impact) and event frequency (amount of dislocation movement and planeslip movements in a certain time frame). Thereby an indirect quadratic proportionalitybetween acoustic emission amplitude and contact time was found.This proportional relationship is also valid for RMS signals with short averagingwindows if system damping is low. For event frequency and RMS signalsthe results of the tensile tests suggest a direct proportional relationship.Furthermore, Hertzian and EHL contact impacts were studied and compared.Thereby it was observed that the overall amplitude of the signal increases forEHL contacts in comparison to Hertzian contacts. In addition the third zerofrequency disappears, which is most likely due to cavitation effects. Furthermore,the results show a shift of the first and second zero frequency towardshigher frequencies, which is caused by the localised deformation of EHL contactsas a result of the solidification of the lubricant. This behaviour of zerofrequencies was in line with simulation results. However, the agreement betweensimulation and measurement for the location of zero frequencies and thesignal amplitude was not satisfying. This mismatch was most likely caused bythe assumption of the global contact force acting at a single point, causing aperfect elastic deformation in the simulation. Additionally, for the findings regardingzero frequencies, a change in the excitation of resonances above thefirst zero frequency in boundary restricted systems was also found, comparingHertzian and EHL impacts.Finally, full scale tests on a complete rolling element bearing were carried outduring the PhD project to validate findings of the single contact experiments.Magnetite contaminated rolling element bearings and their acoustic emissionsignals were investigated with respect to the use of sulfur additives, contaminationand rotational speed. These tests were executed at varying speed forsingle measurements and constant speed for continuous measurement recording. The results of the full scale tests showed good agreement with previousresults of the component tests, such as bouncing ball and tensile tests. Transientforces are the main source of signals for well lubricated rolling elementbearings or bearings at high rotational speed, while acoustic emission signalsof contaminated bearings at low rotational speed were dominated by plasticdeformation signals. Furthermore, it was found that sulfur additives reducethe plastic deformation signal by up to 70% in comparison to contaminatedbearings lubricated with plain grease.
Luleå: Luleå University of Technology, 2016.