Overview of Sensory Data

the case in endurance testing, is still unknown.

A mechanical design was dismissed due to the reasons given in chapter 5.5. The advantages of mechanical solutions are very high speeds and little to no drag in reaction time.

The finally chosen solution is the use of a hydraulic cylinder with voice coil valves.

Voice coil valves are available for a short time on the market. As the name suggests, these valves convert electrical signals to valve positions by means of voice coil actuators [Nas06, SK12]. The advantage of voice coil valves in comparison to servo valves is the higher positioning speed. The working principle of voice coil valves is shown in fig. 5.8.

5.6.2. Bearing Bracket

Concerning the design of the bearing brackets two basic types can be distinguished. The first type of bracket provides a housing for ORs. This type is especially advantageous when standard catalogue bearings are tested, as the brackets can remain within the test rig while only the bearings are replaced. Moreover, the sensitive vibration sensors connected to the brackets do not have to be disconnected. The draw is that lubrication feeding via drilled holes – if required – is connected with considerable effort. Also, the brackets have to be precisely set up and, as chapter 5.10 will show, minimal deviation in the bearing diameters may result in an additional adjustment of the brackets. Figs 5.9 (a) and (b) show a solid bearing bracket, as it has been used in the preliminary tests (chapter 5.9) and a weight optimised design which has been envisioned to increase the dynamic of the actuator.

The second type of bracket integrates the OR of a bearing (fig. 5.9 (c)). The only advantage of this design is to be more convenient for lubrication feeding through drilled holes. A lightweight design as in fig. 5.9 (b) is not practical, since the lids depicted now have to double as flanges for the cage.

The test rig in its current version is capable to handle both types of brackets. The lightweight bracket has not been necessary for the diameters used in the current work.

All brackets are designed to mount vibration sensors for solid-borne sounds.

5.7. Overview of Sensory Data

Solid-borne sound detection Flaking of material from IR, OR or rolling elements generate impacts which excite vibrations. Long before these vibrations are audible to human ears, they can be detected by sensors. The detection quality increases with increasing magnitude of the vibration compared to the general vibration. Therefore these sensors are mounted directly on the OR.

The vibration sensors are linked with a real-time system evaluating the vibration sig-nals and sending stop triggers to the test rig’s control when a damage is detected. To avoid erroneous stop triggers the system evaluates the signals based on a given kinematic model. For this purpose also the torque signal (shaft vibration) and rotation speed is looped into the evaluation system. (appendix E)

CHAPTER 5. DESIGN AND DEVELOPMENT OF A TEST RIG FOR DYNAMIC TESTING UNDER HIGHLY TIME-VARIANT LOAD CHANGES

(a) Type I, solid design (b) Type I, weight optimised design

(c) Type II

fig. 5.9.: Principle design of type I and type II brackets.

5.7. OVERVIEW OF SENSORY DATA

Up to this point the employed system does not differ from common detection systems.

Due to the nature of testing a continuous measurement is not possible, as the dynamic loads will lead to vibrations of higher magnitude and higher orders and thus diminishing the sensitivity of the sensors. In order to mend this problem the test program has intervals of constant loads, during which the test rig’s control activates the monitoring system. Moreover, each test bearing will generate unique magnitudes of vibration levels due to the nature of the testing, the spectrum of which may change during shakedown.

Therefore – unlike in conventional monitoring systems – the first 50 intervals of constant loads are used by the system to learn the specific spectra of each bearing and to create monitoring envelopes of the specific magnitudes in frequency domain (5.10). In the event of failure at a later time the specific damage frequency’s magnitude will rise above the magnitude of the envelope, which now represents the limit of normal operation at the specific frequency.

Signal Frequencies

MagnitudeofSignal

Envelope

fig. 5.10.: The envelope function represents magnitude limits of normal operation over frequencies.

Bearing Force Measurement Regarding the forces acting on each bearing is of special importance. On the one hand, the actual load distribution after each specimen assembly would else be unknown, on the other hand, the change of load distribution with passing time needs to be monitored. Moreover the effects of the system’s oscillation during are not known a-priori.

The existing options for fitting force sensors are force transducers with piezo elements or dynamic strain-gauges. Experience values from other test rigs show that the use of piezo based force transducers are problematic, as piezo element signals experience time dependent drift at higher temperatures and long service times while temperature compensated strain-gauges maintain their signal for considerably longer times. Thus strain-gauges with high dynamic are used in this current test rig.

Also of interest is the stiffness of a bearing during operation. Although not imple-mented at present, provisions are made for measuring the stroke path of the actuator

CHAPTER 5. DESIGN AND DEVELOPMENT OF A TEST RIG FOR DYNAMIC TESTING UNDER HIGHLY TIME-VARIANT LOAD CHANGES

for this purpose.

Oil Temperature Measurement The temperature of the bearing lubrication and cool-ing is measured at the entrance and the exit of the system.

Torque Measurment In order to assess the change of friction over lifetime torque is measured. A torque sensor with a measuring frequency of 44 kHz can also double as a vibration sensor for the shaft at the same time.