rifj static and dynamic specimen stresses were measured over the test range by strain gauging a specimen and using a dead weight calibration to check the stresses calculated from the load cell output. Alignment test results showed that transverse bending amounted to 1% of the applied load and achieved cantilever stresses were 1$ less than for the calibration cantilever static loading. Fatigue test stress levels were set using specimen stress/load cell amplifier voltage data presented in tabular form and calculated from a least squares fit to the calibration information.
The electronic trip circuit which was used is shown in fig. 2.8. The circuit was mechanically triggered by a micro switch. As the specimen failed the increased def lection caused the specimen to strike the micro switch, thus opening the contacts. The principle of operation of the electronic circuit is described in the section dealing with Electromagnetic Vibrator Test Rigs.
1.2 Keelavite Hydraulics - 3? tf Servo Rig
This rig was designed to fatigue specimens machined from strip mild steel in axial tension/compression, and is shown by the photograph in fig. 2.9. Design details are given by drawings ES2/210/1 - 2. The framework was fab ricated from structural steel sections and the load cell fastened to the crossbar of the framework. The ram was bolted to the base of the rig. The specimen was held in grips between the load cell and the ram (see fig. 2.10) and, to prevent buckling in compression, was loaded dynamic ally about a mean tensile load of 800 lbf. Alignment of the hydraulic ram and load cell to give accurate axial
- Ill* -
loading was achieved by shimming and using the mandrel and collar system shown in fig. 2.11. The mandrel was quite free when the load cell and ram had finally been bolted in position.
One modification to the electronic control module had to be made during rig commissioning. Electrical noise, mainly from the switching transients of an overhead crane and workshop machinery, was picked up by the load cell and associated cable and amplified by the load cell amplifier. This interference was not noticeable when amplifier common mode balance had been restored by achieving reactive and resistive balance in the amplifier feedback circuits (see Appendix a) •
Loading pulses could also be felt on the specimen by sv/itching electrical equipment on/off the mains supply which powered the control module. To avoid thi3 trouble, the module mains power was taken from a separate clean mains circuit.
Further modifications were made to the electronic control circuit to improve the rig frequency response as described in Appendix A, which also gives the gains of the components within the loop. Details of the control circuit electronics are given in fig. 2.12. Drift measurements of the load cell amplifier output are shown in fig. 2.13, giving an apparent mean load variation of 15 lbf> or l££ of the set mean level. This was acceptable for the testing here. The test rig frequency response i6 shown in fig. 2.1b*
To maintain a high degree of axial alignment both ends of test specimens were first located in each grip by a dowel which was a slide fit through brass bushes in the grip and a reamed hole in the specimen. The eccentricity of the reamed
hole from the specimen test section was not greater than 0.005". Then the 800 lbf. mean tensile load v,as applied and the specimen clamped by bolts, tightened using two spianners to minimise any twisting effect. All testing was carried out at a mean load of 800 lbf.
The load cell/load cell amplifier calibration was made using the 5 Tf. range of an Amsler b0 Tf. testing machine. Achieved test rig static and dynamic specimen stresses were measured over the test range by strain gauging b specimens and using dead weight calibrations to check the stresses calculated from the load cell output. The results are summarised in Table 2.1.
A factor complicating any rational analysis of these results was the condition of the specimens themselves. Since the material was supplied in the hot rolled condition it was doubted if any one specimen was perfectly flat. During clamping in the grips the straightening of a curved specimen would induce transverse bending stresses along its length and in-plane bending stresses along its width. The alignment test results indicated that at the worst there might be a variation in the mean tensile stress across the specimen of about 20 % from the nominal applied, and that the dynamic stress across the specimen could vary by up to about
8 % . It is questionable whether the order of variation of static mean stress indicated by these tests would sig nificantly affect the results. Work with sheet mild steel specimens by Frost & Greenan (2.0) has indicated that for mild steel the mean load has no effect on cracsc propagation rate. The fatigue results showed no more scatter than that expected from polished specimens, and out of 70 specimens tested, 21 fractured from one side, 23 from the other and
26 specimens fractured from cracks initiated from both