Fatigue Design and Evaluation Committee |
FD&E HomeSearchEvents CalendarShort CoursesMinutes, Presentations, AgendasContact InfoATV ProjectWeld ProjectBolt ProjectDivisionsData OnlineLinksSteering Commitee
|
Unconfirmed Minutes - Fatigue Life Prediction DivisionSAE Fatigue Design and Evaluation Committee October 17 and 18, 1995 Toronto, Ontario Divisional Progress Meeting (October 17):The meeting was called to order and the minutes from the March meeting in LaConner, Washington were approved as written.Rolland Hartline from Clark Engineering Services presented his fatigue analysis of the Honda ATV data. He also reported the variability of fatigue life predictions made from strain measurements on construction equipment. For the Honda ATV data, he separated the pallet impacts from the travel after impacts. Performing a rainflow fatigue analysis, he calculated cumulative damage for each rainflow range in the load history, using component load-life damage curves with slopes of -0.1 and -0.33 respectively. The pallet impact data were normalized to a test length of 300 hours. He then compared the relative severity of the various events in terms of the rainflow ranges. For the -0.33 slope, the largest events contributed much less to the total life than for the -0.10 slope analysis. The relative damage contribution of the various rainflow ranges was very sensitive to the slope of the damage curve used for the analysis. In the second part of his presentation, Rolland showed the variability of fatigue life, calculated from strains measured on equipment under service loads, at various test site locations. Calculated damage varied by a factor of ten for the different test locations. Presentation slides are attached. Darrell Socie from the University of Illinois presented "Statistical Fatigue Analysis". He separated a fatigue analysis into the following areas: loading history, material properties, manufacturing (geometry and processing), and analysis methodology. In his analysis, he created a cumulative exceedence diagram of rainflow ranges of a load history and then modeled it using a Weibul distribution and a coefficient of variation of 10%. He also applied a coefficient of variation of 10% to the material properties and to the notch factor. A Monte Carlo simulation was performed to calculate the distribution of resulting fatigue lives. Correlation coefficients were calculated for the input variables to track their relative influence on the variability of the calculated lives. Nearly half of the calculated life variation was attributed to the load history alone. Prof. Socie concluded his presentation by showing a plot of calculated life /experimental life vs. standard deviation for the SAE Keyhole program. The result had nearly the same slope as in the previous example analysis, leaving everyone to wonder about the variation in fatigue analysis methodology. Presentation slides are attached. Art Page was unable to attend the meeting Steve Tipton from the University of Tulsa presented his ideas on developing notched shaft design algorithms to handle stress concentrations. The algorithms would contain elastic solutions, equivalent stress-strain relationships, and other valid relationships such as constant stress ratio, constant energy, etc. that allow the determination of a solution. He cited the techniques that currently exist for blunt circumferential notches and suggested that similar methods could be used to cover other notch geometries such as splines, keyways, and sharp notches. Each algorithm would cover a valid geometry and would contain the appropriate solutions and simplifying relationships for that particular type of geometric feature. Presentation slides are attached. Mark Barkey from the University of Alabama made a presentation entitled "Fatigue Life Calculations using Proportional Notch Estimation Rules". Using the proportional load data from the SAE notched shaft program, reported in AE14, he compares constant stress and constant strain ratio approaches in the notch calculations. Neuber's rule and the Seeger modification to Neuber's rule were used. Fatigue life calculations were compared to the experimental lives reported in AE14. He concludes that the pseudo stress and constant strain ratio approaches result in the same equivalent strain. The largest difference in the two approaches lies in the choice of a notch curve (Neuber or Seeger). Divisional Planning Meeting (October 18):Discussions were focused on the following four areas.Surface Finish DataThe surface finish fatigue data obtained by the Component Testing Division was discussed. Life prediction work to evaluate the relative effects of residual stresses and geometric notches was encouraged. Ideas included separating geometry and residual stress effects, comparing strain life to fracture mechanics approaches, and developing fatigue notch factors for the various surface and hardness conditions. Please refer to the data published by the Component Testing Division in the minutes from the previous three meetings and/or contact Phil Dindinger.Powder Metals TestingLife Prediction Division involvement in the Powder Metal Task Group was discussed. We are looking at places where life predictions would complement and extend the work on powder metals. Data generated by this study could be used to develop and evaluate life prediction algorithms, etc. Strain life vs. Fracture Mechanics was discussed. Contact Steve Tipton and Ralph Stephens for more information on the work and testing being performed by the Powder Metals Task Group.SAE Hardened ShaftThe idea of making carefully controlled rosette strain measurements on a soft notched shaft under a discriminating multiaxial load path was discussed. Other shafts would be tested to cracking under the same load path to generate fatigue data. It was felt that these data would be useful in furthering our understanding of current multiaxial fatigue life prediction methods. Defining the load path would be central to this effort. It would be chosen to discriminate between many of the methods currently used. This work is in the proposal/planning stages; contact Tom Cordes with your ideas or questions.StatisticsThe subject did not generate any discussion.Respectfully submitted, Michael Messman |
