Informazioni sul documento
- Università
- Politecnico di Milano
- Corso di laurea
- Mechanical Engineering
- Materia
- Mechanical Systems Dynamics
- Anno accademico
- 2025-2026
- Classificazione
- Altro materiale
- Formato originale
- Testo
- Testo ricercabile
Materiale universitario di Mechanical Systems Dynamics per il corso di Mechanical Engineering presso Politecnico di Milano. Il documento tratta: Advanced Dynamics of Mechanical Systems Assignments 1 & 2 Experimental Modal Analysis & Finite Element Modelling Alper Belen Prof. Stefano Melzi – Dott. Giulia Pomaranzi Politecnico di Milano – A.Y. 2025/2026 June 2026 Abstract This report addresses Part A of Assignment 1, which…
Materiale universitario di Mechanical Systems Dynamics per il corso di Mechanical Engineering presso Politecnico di Milano. Il documento tratta: Advanced Dynamics of Mechanical Systems Assignments 1 & 2 Experimental Modal Analysis & Finite Element Modelling Alper Belen Prof. Stefano Melzi – Dott. Giulia Pomaranzi Politecnico di Milano – A.Y. 2025/2026 June 2026 Abstract This report addresses Part A of Assignment 1, which…
Qualità dell’importazione: il testo è stato estratto direttamente dal documento originale.
Passaggi rappresentativi riconosciuti nelle diverse parti del materiale. Il testo completo resta presente nella pagina per la ricerca, mentre l’anteprima compatta rende più semplice la lettura.
Advanced Dynamics of Mechanical Systems Assignments 1 & 2 Experimental Modal Analysis & Finite Element Modelling Alper Belen Prof. Stefano Melzi – Dott. Giulia Pomaranzi Politecnico di Milano – A.Y. 2025/2026 June 2026 Abstract This report addresses Part A of Assignment 1, which applies an FRF based multi mode curve fitting identification procedure to a an aluminium cantilever beam of rectangular cross section which is analysed analytically and numerically to obtain its natural frequencies and bending mode shapes from the standing wave solution of a slender beam. A set of Frequency Response Functions (FRFs) is then computed for a fixed excitation point and several measurement positions; these numerical FRFs are then treated as synthetic experimental data. The modal parameters: natural frequencies, damping ratios and mode shapes of the first four modes are recovered through a nonlinear least-squares curve fitting performed simultaneously on all FRFs of the structure. The identified parameters are compared with the values used to generate the data. Because the synthetic FRFs are free of noise and the modes are well separated, the identification reproduces the original parameters to within a few parts in ten thousand, confirming the correctness of the implementation. 1 Introduction Experimental Modal Analysis (EMA) extracts a structure’s modal parameters natural frequencies, damping factors and mode shapes from measured response data, with the aim of either updating or validating a numerical model. Part A of the assignment provides a controlled environment for this procedure: rather than starting from physical measurements, the “experimental” FRFs are generated numerically from a known beam model. This has two advantages. First, the exact answer is available so the…
3.5 Comparison of simulated and identified mode shapes The mode shapes recovered from the identified residues are compared with the analytical mode shapes in Figure 5. For each mode the continuous analytical curve, the analytical values sampled at the six sensor positions, and the identified values are shown together after a common normalisation. Simulated vs identified mode shapes Mode 1: 4.50 Hz Mode 2: 28.22 Hz 1.0 Model (continuous) 1.0 Model at points Identified 0.8 0.5 0.6 norm. norm. 0.4 0.0 0.2 0.5 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 x [m] x [m] Mode 3: 79.03 Hz Mode 4: 154.87 Hz 1.0 1.0 0.5 0.5 norm. norm. 0.0 0.0 0.5 0.5 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 x [m] x [m] Figure 5 – Comparison between the analytical mode shapes (continuous line and square markers at the sensor positions) and the mode shapes identified from the residues (circular markers). The identified points fall almost exactly on the analytical mode shapes for all four modes, confirming that the residues obtained from the multi-mode curve fit carry the correct spatial information. The comparison is necessarily limited to the discrete sensor locations, which is intrinsic to the modal-residue approach; with six well-distributed sensors, however, the sampled shapes already capture the essential features of each mode, including the sign reversals at the internal nodes. 3.6 FRF reconstruction As a final consistency check, the FRF of the reference output point is reconstructed by summing the contributions of all four identified modes (each within its own band) and compared with the original numerical FRF in Figure 6. 8
5.5.2 Mode 4 Mode 4, at ffull = 0.536 Hz, lies in the crowded region of the spectrum where modes 2, 3 and 4 are closely spaced. As a consequence the averaged FRF (Fig. 12) is noisier and the peak is broader, so the damping estimate is somewhat less reliable than for mode 1; nevertheless the two estimators still agree reasonably well (h4 ≈ 0.019–0.022). The reconstructed mode shape (Fig. 14) is in good agreement with the fourth FE mode, confirming a correct mode pairing. Mode 4 – Averaged FRF | #HP=0.022, #PD=0.019 #10 -3 f0=4.480 Hz 10 |FRFavg| 9 f0 |FRF| (m/N) Half-power 8 1/#2 level 7 6 4.3 4.35 4.4 4.45 4.5 4.55 4.6 4.65 Frequency (Hz) Phase of averaged FRF 150 Phase (°) 100 50 4.3 4.35 4.4 4.45 4.5 4.55 4.6 4.65 Frequency (Hz) Figure 12 – Mode 4 – averaged FRF (amplitude and phase). The broad peak reflects the influence of the neighbouring modes. 15
flexible) beam family evaluated at the top of the analysis band, v s u u π2 EJ Lmax = , (16) t η Ωmax m where Ωmax = 2πfmax is the highest circular frequency of interest and η is a safety factor. With an analysis band of 0–6 Hz (fmax = 6 Hz) and η = 1.5, evaluating (16) for the three beam families gives the upper bounds in Table 6. The governing (smallest) value is Lmax ≈ 11.6 m for the green beams; the element length used in the .inp file, ≈ 1 m, lies well below this bound, so the mesh resolves every mode in the band with margin. Table 6 – Maximum admissible element length per beam family (fmax = 6 Hz, η = 1.5). Beam family Lmax [m] Red beams 14.9 Green beams 11.6 Blue beams 23.5 7.2 Finite-element model and constraints The global mass and stiffness matrices M and K are assembled element by element from the per-element properties (mass per unit length m, axial stiffness EA and bending stiffness EJ) read from the input file. With nnod nodes the assembled matrices have size 3 nnod , since every planar node carries three degrees of freedom (DOF): the horizontal and vertical translations and the in-plane rotation. The vertical spring at B is introduced by adding its stiffness to the diagonal entry associated with the vertical DOF of B, KvB vB ← KvB vB + k. (17) The DOFs are then partitioned into a free set (F ) and a constrained set (C), the latter collecting the DOFs suppressed by the supports at the base of the two towers. Re-ordering rows and columns accordingly, each global matrix takes the block form " # " # MF F MF C KF F KF C M= , K= , (18) MCF MCC KCF KCC and the same partition is applied to the damping matrix C. The subscript F F denotes the block coupling free DOFs to free DOFs, F C (with its transpose CF ) the coupling between free and constrained DOFs, and CC…
#10-6 Vertical displacement at B — moving jumper 5 jumper leaves track 0 v B(t) [m] -5 -10 -15 0 5 10 15 20 Time [s] Figure 24 – Vertical displacement at B produced by the jumper moving from A to B; the dashed line marks the instant the jumper leaves the track. 9 Conclusion Assignment 2 applied the FEM to dynamic analysis of a ski jump model we have. After assembling the model and adding the elastic support at B, the first four natural frequencies and mode shapes were obtained from an undamped eigenvalue problem. The modal damping is obtained by a Rayleigh model. The FRFs computed by direct inversion exhibit resonances exactly at the identified natural frequencies, and the comparison between the full and the two mode modal solutions confirmed that a static residual correction is essential to recover the low-frequency response and the anti-resonances of a truncated modal model. The static analysis located the region of maximum deflection under the snow load along the slender overhang, while the optional moving load simulation showed a pronounced transient response at B during the jumper’s transit followed by a lightly damped free vibration. Overall, the study illustrates how a relatively small FE model captures both the modal properties and the static and transient response of the structure with good consistency between the different solution strategies. 29
Prima pagina del documento.