This report documents the fatigue code SIESTA that has been used for recently here at Sandia National Laboratories. It is written in two parts: the first as a user manual and the second as a theory manual. Currently employed in SIESTA are stress-life cycle approaches. Clients have requested the use of standards in particular analyses; therefore, the American Society of Mechanical Engineers, Boiler Pressure Vessel Code fatigue standards have been implemented. These include an elastic, an elastic-plastic, and a weld fatigue method. All three methods use a Max-Min cycle counting method that is appropriate for non-proportional loading. A Signed von Mises method that used a Rainflow Cycle Counting Method is also implemented. The Signed von Mises with the Rainflow Cycle Counting Method is appropriate for proportional loading. Several verification examples are noted and include comparisons to experimental data.
This work explores deriving transmissibility functions for a missile from a measured location at the base of the fairing to a desired location within the payload. A pressure on the outside of the fairing and the rocket motor’s excitation creates an acceleration at a measured location and a desired location. Typically, the desired location is not measured. In fact, it is typical that the payload may change, but measured acceleration at the base of the fairing is generally similar to previous test flights. Given this knowledge, it is desired to use a finite-element model to create a transmissibility function which relates acceleration from the previous test flight’s measured location at the base of the fairing to acceleration at a location in the new payload. Four methods are explored for deriving this transmissibility, with the goal of finding an appropriate transmissibility when both the pressure and rocket motor excitation are equally present. These methods are assessed using transient results from a simple example problem, and it is found that one of the methods gives good agreement with the transient results for the full range of loads considered.
Motivation: Crucial aspect of mechanical design is joining methodology of parts. Ability to analyze joint and fasteners in system for structural integrity is fundamental. Different modeling representations of fasteners include spring, beam, and solid elements. Various methods compared for linear system to decide method appropriate for design study. New method for modeling fastener joint is explored from full system perspective. Analysis results match well with published experimental data for new method.
There are several methodologies for modeling fasteners in finite element analyses. This work examines the effect of four predominant fastener modeling methods regarding the fatigue of mock hardware that requires fasteners. Typical fastener modeling methods explored in this work consist of a spring method with no preload, a beam method with no preload, a beam method with a preload, and a solid model representation of the fastener with preload. It is found that the different fastener modeling methods produce slightly different fatigue damage predictions, and that this uncertainty in modeling is insignificant as compared to uncertainty in input. Consequently, any of these methods are considered appropriate. In order to make this assertion, multiaxial fatigue methods are investigated and a proportional method is used because of a biaxiality metric.
The goal of this work is to build model credibility of a structural dynamics model by comparing simulated responses to measured responses in random vibration environments, with limited knowledge of the true test input. Oftentimes off-axis excitations can be introduced during single axis vibration testing in the laboratory due to shaker or test fixture dynamics and interface variation. Model credibility cannot be improved by comparing predicted responses to measured responses with unknown excitation profiles. In the absence of sufficient time domain response measurements, the true multi-degree-of-freedom input cannot be exactly characterized for a fair comparison between the model and experiment. Methods exist, however, to estimate multi-degree-of-freedom (MDOF) inputs required to replicate field test data in the laboratory Ross et al.: 6-DOF Shaker Test Input Derivation from Field Test. In: Proceedings of the 35th IMAC, A Conference and Exposition on Structural Dynamics, Bethel (2017). This work focuses on utilizing one of these methods to approximately characterize the off-axis excitation present during laboratory random vibration testing. The method selects a sub-set of the experimental output spectral density matrix, in combination with the system transmissibility matrix, to estimate the input spectral density matrix required to drive the selected measurement responses. Using the estimated multi-degree-of-freedom input generated from this method, the error between simulated predictions and measured responses was significantly reduced across the frequency range of interest, compared to the error computed between experimental data to simulated responses generated assuming single axis excitation.