Finite Element Model Calibration for Ares I-X Flight Vehicle

Abstract

Ares I-X is a flight test vehicle developed by NASA to demonstrate a new class of crew launch vehicle. For this first flight test, the first stage was a four segment solid rocket booster with mass simulators used to represent the other sections of the Ares I vehicle. Although this vehicle is significantly simpler than the Ares I, model calibration was required for the finite element model used in loads analysis and flight control evaluations before its maiden flight. The process of calibrating models involves updating parameters and reconciling predictions with test data. This work presents a probabilistic approach to the calibration process. The approach uses Analysis of Variance (ANOVA) for parameter sensitivity, nonlinear optimization to minimize the error between test and analysis, and multiple FEM models to bound the system response and to assess the probability of finding a reconciling solution. To reduce the computational burden associated with ANOVA, response surface models are used in lieu of computationally intensive finite element solutions. Uncertainty in the parameters and their effect on the frequency response function is studied in terms of Principal Values of the frequency response functions. Uncertainty bounds of the principal values are established across multiple models to allow one to determine the probability of finding a solution that reconciles analysis with test results. Results from applying this model calibration process to the Ares I-X project are described. Findings presented in the paper confirmed that the baseline model used for pre-flight assessments was within the acceptable range established for guidance and control.

Using Dispersed Modes During Model Correlation

“…As described by Horta, et al, 5 Aft Skirt (stiffeners and shell), Forward Skirt Extension, and Frustum. Modulus of elasticity is allowed to vary by ±10%, while USS joint stiffness is allowed to vary by ±50%.…”

“…For the Ares I-X flight model dispersions, only material and spring properties used for model correlation by Horta, et al 5 are varied during the dispersion process. The material properties include the Young's modulus for the LAS, Upper Stage Simulator, First Stage, Aft Skirt Stiffener, Forward Skirt Ext and Frustrum, XL and Forward Skirt, and Aft Skirt Shell.…”

“…For modal testing, the Ares I-X flight vehicle was placed on top of the mobile launcher, supported by the four hold-down posts, and excited using 4 hydraulic actuators. From the resulting test data, frequency, damping, and mode shapes for all target modes below 12.5 Hz are estimated as described by Horta, et al 5 2177 dispersions for the Ares I-X flight vehicle are plotted using the same error criteria and Euclidean norm discussed above and shown in Fig. 14.…”