ECL15_Damping_SDTools
0 likes268 views
This document summarizes design methods for damping augmentation and integrating damping models into global simulations. It discusses how insufficient elastic coupling is often the main reason damping design fails. It advocates analyzing flexibility, contact, and sliding in junctions to ensure sufficient coupling. Sample mechanisms are presented that can be optimized with parameters. Integration of damping models into reduction tools, transient and frequency solvers, and post-processing tools is covered. The importance of understanding trends through analysis tools is also highlighted.
ECL15_Damping_SDTools
- 1. E. Balmes, G. Vermot des Roches SDTools, Arts et Metiers ParisTech Workshop “Amortissement”. ECL, Oct 8, 2015 1 Design methods for damping augmentation & integration of models in global simulations. Industrial illustrations with SDT
- 2. • Meshing • Material handling • Response & pole predictions • Vibroacoustic optimization • Analysis tools Damping design process
- 3. Outline • Main reason for damping design failure : missing elastic design – Level of coupling : a necessity for damping – Meaning in junctions analyze flexibility, contact, sliding – Sample mechanisms with parameters that can be optimized • Integration in global simulations / analysis tools – Reduction tools & base damping – Direct transient & frequency solvers – Post-processing tools – Force/state map in physical and macro-models 3
- 4. Apparent damping with pure contact 4 Thénint 2011
- 5. Damping occurs in the transition 5 • Contact with broadband excitation level • Viscoelasticity : temperature/frequency dependence • Friction : amplitude/normal force dependence ECL/Laxalde 2007 ENSAM/Hammami 2014
- 6. Low joint energy no damping Simple example : bending joint Design space : k=kv(1+ih) with kv variable, h=1 Output : f(kv,h) z(kv,h) Damping can only exist if joint is « working » Energy fraction in joint must be « sufficient » (depends on kv) Soft joint Stiff joint Dissipative joint 1 2 3 1 2 3 Mode2 Mode1 cvwj=kvh kv Similar ideas : electromecanical coupling coefficient, pole/zero distance, modal strain energy
- 7. SUPMECAvisco Washer ENSAM Contact/friction & coupling 7 • Variable contact surface, contact, sliding can generate coupling Damping design • Phase 1 : sufficient coupling (elastic model) • Phase 2 : device optimization (physical damping model) Chassis Brakes International PSA
- 8. Damped viscoelastic junction Initial design : bolted brackets • Non uniform pressure • Friction potential limited Energy analysis : full visco-layer • Energy in screw decreases rapidly • There is an optimal visco layer stiffness Structured layer : • Carry load : washer • Damp : visco 8Hammami/Balmes/Guskov, MSSP 2015 Washervisco *3 *8
- 9. Outline • Main reason for damping design failure : missing elastic design – Level of coupling : a necessity for damping – Meaning in junctions analyze flexibility, contact, sliding – Sample mechanisms with parameters that can be optimized • Integration in global simulations / analysis tools – Reduction tools & base damping – Direct transient & frequency solvers – Post-processing tools – Force/state map in physical and macro-models 9
- 10. Modeling damping mechanisms Models : known • Visco • Point to point connection : force-state map F = f(q,t) • Surface contact/friction model Difficulties • Need for parametric studies – Viscoelasticity : temperature/frequency dependence – Friction : amplitude/normal force dependence Solver cost & robustness • Need to understand trends Analysis tools Laxalde 2007
- 11. Dimensions of damping design problems • Model parameters (designs, dispersion) • Operating point (temperature, pressure) • “Frequency” domain : linearization, HBM, NL modes (frequency, level) • Time domain transients (loading, initial cond., …) Needed : global solution • Solver optimization • Reduction : realistic FEM, parametric problems • Analysis tools : “features” similar to modes 11 Nom . +10 % +20 % -20%
- 12. Solver optimization Most difficult to obtain significant gains SDTools activities • Extensible framework for NL description • Transient : implicit : fixed jacobian penalized contact, theta method (AcouFren), explicit : SE generation (MAIAS) • Frequency : parametric viscoelastic studies, linearization along transient, HBM (starting in CLIMA) • Modal damping in full FEM • Fine // of residue computation • Coarse // : bypass cost problem by distribution of parametric computations 12
- 13. Reduction : easiest gains • Craig-Bampton often sub-performant because of interfaces • CMT (component mode tuning) – Exact nominal modes (possibly multi-model) – Replaceable unreduced NL area – Possibly modal component coordinates • Error control & SVD • Compression of output data 13 𝜙 𝐹𝑢𝑙𝑙 𝐴𝑟𝑒𝑎 𝐼 PhD : Bobillot (ADS), Sternchuss (SNECMA), Vermot (Bosch), Thénint (EDF)
- 14. Analysis tools : modal features • Modal & mechanical energies • Instant mode computation Sample question : does the modeshape change in a NL response ? Ep(t) Ek(t) 14
- 15. 15 Guillaume Vermot des Roches - Design oriented simulation of squeal instabilities - 27 / 01 / 2011 Component Mode Sensitivities • Identification of sensitive parameters:
- 16. F = f(q,t) : force state maps Classical rheology 𝜎 = 𝑓 𝜖 𝑡 =local in space Classical macro-model : • Model simplification • Arbitrary selection of q • Equivalent building of point f Physical macro-model : • Junctions are distributed • What is the state q ? • How is f field distributed ? Macro-model = force/state map 16
- 17. Analysis tools : macro-forces • Principal loads 𝐹 𝑛𝑙 𝑞 𝑠 , 𝑘 𝑛, 𝑘 𝑡 𝑁×1 ≈ 𝑘 𝑛 KN 𝑉𝑗 𝑎𝑗 KN 𝑉𝑗 𝑇𝑁 𝑁 𝑗=1 𝑞(𝑠) 𝑁×1 • Dissipation localization 17
- 18. 50 100 150 200 250 300 10 -4 10 -3 10 -2 10 -1 In:Guide X, Out:FLOOR-Z Frequency FRF Conclusion • Elastic design must be understood • Models & solvers – Accessible with state of the art methods – Industrially useful only when parametric studies and physical models are accessible – The focus should be on design process & robustness (temp, pressure, variability) • Feature extraction strategies are fundamental – System & component mode characteristics – Macro-forces – Performance targets www.sdtools.com/Publications.html 18 RestartIdentification Level