INTERVAL ARITHMETIC APPLIED TO STRUCTURAL DESIGN OF UNCERTAIN MECHANICAL SYSTEMS

Σχετικά έγγραφα
DETERMINATION OF DYNAMIC CHARACTERISTICS OF A 2DOF SYSTEM. by Zoran VARGA, Ms.C.E.

D Alembert s Solution to the Wave Equation

Exercises 10. Find a fundamental matrix of the given system of equations. Also find the fundamental matrix Φ(t) satisfying Φ(0) = I. 1.

Numerical Analysis FMN011

Operational Programme Education and Lifelong Learning. Continuing Education Programme for updating Knowledge of University Graduates:

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

Chapter 6: Systems of Linear Differential. be continuous functions on the interval

New bounds for spherical two-distance sets and equiangular lines

Ingenieurbüro Frank Blasek - Beratender Ingenieur Am Kohlhof 10, Osterholz-Scharmbeck Tel: 04791/ Fax: 04791/

Ingenieurbüro Frank Blasek - Beratender Ingenieur Am Kohlhof 10, Osterholz-Scharmbeck Tel: 04791/ Fax: 04791/

g-selberg integrals MV Conjecture An A 2 Selberg integral Summary Long Live the King Ole Warnaar Department of Mathematics Long Live the King

Eulerian Simulation of Large Deformations

General 2 2 PT -Symmetric Matrices and Jordan Blocks 1

Partial Differential Equations in Biology The boundary element method. March 26, 2013

Example Sheet 3 Solutions

The ε-pseudospectrum of a Matrix

Biostatistics for Health Sciences Review Sheet

Low Frequency Plasma Conductivity in the Average-Atom Approximation

The wave equation in elastodynamic

Constitutive Relations in Chiral Media

DiracDelta. Notations. Primary definition. Specific values. General characteristics. Traditional name. Traditional notation

A Lambda Model Characterizing Computational Behaviours of Terms

ES440/ES911: CFD. Chapter 5. Solution of Linear Equation Systems


6.3 Forecasting ARMA processes

ExpIntegralE. Notations. Primary definition. Specific values. Traditional name. Traditional notation. Mathematica StandardForm notation

Απόκριση σε Μοναδιαία Ωστική Δύναμη (Unit Impulse) Απόκριση σε Δυνάμεις Αυθαίρετα Μεταβαλλόμενες με το Χρόνο. Απόστολος Σ.

Concrete Mathematics Exercises from 30 September 2016

A Bonus-Malus System as a Markov Set-Chain. Małgorzata Niemiec Warsaw School of Economics Institute of Econometrics

ΕΠΙΧΕΙΡΗΣΙΑΚΟ ΠΡΟΓΡΑΜΜΑ «ΕΚΠΑΙΔΕΥΣΗ ΚΑΙ ΔΙΑ ΒΙΟΥ ΜΑΘΗΣΗ»

Fundamentals of Signals, Systems and Filtering

Eight Examples of Linear and Nonlinear Least Squares

DERIVATION OF MILES EQUATION FOR AN APPLIED FORCE Revision C

Written Examination. Antennas and Propagation (AA ) April 26, 2017.

2.019 Design of Ocean Systems. Lecture 6. Seakeeping (II) February 21, 2011

Probabilistic Approach to Robust Optimization

Statistical analysis of extreme events in a nonstationary context via a Bayesian framework. Case study with peak-over-threshold data

Statistics 104: Quantitative Methods for Economics Formula and Theorem Review

ECE Spring Prof. David R. Jackson ECE Dept. Notes 2

SCHOOL OF MATHEMATICAL SCIENCES G11LMA Linear Mathematics Examination Solutions

Lecture 21: Properties and robustness of LSE

Tridiagonal matrices. Gérard MEURANT. October, 2008

ECE 308 SIGNALS AND SYSTEMS FALL 2017 Answers to selected problems on prior years examinations

High order interpolation function for surface contact problem

Chapter 1 Introduction to Observational Studies Part 2 Cross-Sectional Selection Bias Adjustment

Review Test 3. MULTIPLE CHOICE. Choose the one alternative that best completes the statement or answers the question.

ΑΝΩΤΑΤΟ ΤΕΧΝΟΛΟΓΙΚΟ ΕΚΠΑΙΔΕΥΤΙΚΟ ΙΔΡΥΜΑ ΠΕΙΡΑΙΑ ΣΧΟΛΗ ΤΕΧΝΟΛΟΓΙΚΩΝ ΕΦΑΡΜΟΓΩΝ ΤΜΗΜΑ ΠΟΛΙΤΙΚΩΝ ΔΟΜΙΚΩΝ ΕΡΓΩΝ

Inverse trigonometric functions & General Solution of Trigonometric Equations

Second Order RLC Filters

Figure A.2: MPC and MPCP Age Profiles (estimating ρ, ρ = 2, φ = 0.03)..

Computing the Macdonald function for complex orders

Homework 3 Solutions

ΗΜΥ 220: ΣΗΜΑΤΑ ΚΑΙ ΣΥΣΤΗΜΑΤΑ Ι Ακαδημαϊκό έτος Εαρινό Εξάμηνο Κατ οίκον εργασία αρ. 2

Global nonlinear stability of steady solutions of the 3-D incompressible Euler equations with helical symmetry and with no swirl

Exercises to Statistics of Material Fatigue No. 5

ΕΙΣΑΓΩΓΗ ΣΤΗ ΣΤΑΤΙΣΤΙΚΗ ΑΝΑΛΥΣΗ

Fourier Series. MATH 211, Calculus II. J. Robert Buchanan. Spring Department of Mathematics

Dr. D. Dinev, Department of Structural Mechanics, UACEG

ECE598: Information-theoretic methods in high-dimensional statistics Spring 2016

Areas and Lengths in Polar Coordinates

ΣΤΑΤΙΚΗ ΜΗ ΓΡΑΜΜΙΚΗ ΑΝΑΛΥΣΗ ΚΑΛΩ ΙΩΤΩΝ ΚΑΤΑΣΚΕΥΩΝ

Notations. Primary definition. Specific values. General characteristics. Traditional name. Traditional notation. Mathematica StandardForm notation

Mellin transforms and asymptotics: Harmonic sums

ECE 468: Digital Image Processing. Lecture 8

ER-Tree (Extended R*-Tree)

Optimal Parameter in Hermitian and Skew-Hermitian Splitting Method for Certain Two-by-Two Block Matrices

The Negative Neumann Eigenvalues of Second Order Differential Equation with Two Turning Points

Areas and Lengths in Polar Coordinates

Lecture 34 Bootstrap confidence intervals

CHAPTER 48 APPLICATIONS OF MATRICES AND DETERMINANTS

A Method of Trajectory Tracking Control for Nonminimum Phase Continuous Time Systems

Problem Set 3: Solutions

Arithmetical applications of lagrangian interpolation. Tanguy Rivoal. Institut Fourier CNRS and Université de Grenoble 1

ΣΥΓΚΡΙΣΗ ΑΝΑΛΥΤΙΚΩΝ ΚΑΙ ΑΡΙΘΜΗΤΙΚΩΝ ΜΕΘΟ ΩΝ ΓΙΑ ΤΗ

Discretization of Generalized Convection-Diffusion

w o = R 1 p. (1) R = p =. = 1

Parametrized Surfaces

Local Approximation with Kernels

Forced Pendulum Numerical approach

ΑΡΙΣΤΟΤΕΛΕΙΟ ΠΑΝΕΠΙΣΤΗΜΙΟ ΘΕΣΣΑΛΟΝΙΚΗΣ

The Simply Typed Lambda Calculus

Reminders: linear functions

Abstract Storage Devices

Introduction to Theory of. Elasticity. Kengo Nakajima Summer

IV. ANHANG 179. Anhang 178

Audio Engineering Society. Convention Paper. Presented at the 120th Convention 2006 May Paris, France

Phys460.nb Solution for the t-dependent Schrodinger s equation How did we find the solution? (not required)


Laplace Expansion. Peter McCullagh. WHOA-PSI, St Louis August, Department of Statistics University of Chicago

Sampling Basics (1B) Young Won Lim 9/21/13

k A = [k, k]( )[a 1, a 2 ] = [ka 1,ka 2 ] 4For the division of two intervals of confidence in R +

Sequent Calculi for the Modal µ-calculus over S5. Luca Alberucci, University of Berne. Logic Colloquium Berne, July 4th 2008

Uniform Convergence of Fourier Series Michael Taylor

CHAPTER 25 SOLVING EQUATIONS BY ITERATIVE METHODS

Outline Analog Communications. Lecture 05 Angle Modulation. Instantaneous Frequency and Frequency Deviation. Angle Modulation. Pierluigi SALVO ROSSI

Thermistor (NTC /PTC)

ΜΑΣ 473/673: Μέθοδοι Πεπερασμένων Στοιχείων

3+1 Splitting of the Generalized Harmonic Equations

If we restrict the domain of y = sin x to [ π, π ], the restrict function. y = sin x, π 2 x π 2

Buried Markov Model Pairwise

Transcript:

INTERVAL ARITHMETIC APPLIED TO STRUCTURAL DESIGN OF UNCERTAIN MECHANICAL SYSTEMS O. Dessombz, F.Thouverez J-P. Laîné, L.Jézéquel Laboratoire de Tribologie et Dynamique des Systèmes UMR CNRS 5513 École Centrale de Lyon 69134 Ecully France

Uncertainties in mechanical systems 2 Modeling: Finite Element Method Hypothesis Deterministic Structure Uncertain Data Variable Variations of the parameters, the geometry Variations of the results - constraints - displacements Computations made for the mean value of the parameters - eigenfrequencies Different types of uncertainties - stochastic parameters: Young s modulus, density, thickness... - unknown parameters : boundary conditions, assembly... - evolutive parameters : aging, variation in time - model uncertainties: behavior law, choice of the mesh... Modeling - stochastic variables generic laws (Gaussian, uniform) particular laws (bounded domain) - Intervals uncertain and bounded variables

Finite Element Problems 3 Γ Formulation : PDE u M Ω L(u) on Ω C(u) on Γ Stationarity of an integral functional J(u) R(u)dΩ δj(u) u E u E u space of the admissible fields Discretization of E u sub domains (FE) and interpolation û nodes U i N i (M) node

Finite Element Problems 4 Elementary formulation Assembly global formulation U1 U2 U3 U4 - Static problems KU F K EI L 3 12 6L 12 6L 6L 4L 2 6L 2L 2 12 6L 12 6L 6L 2L 2 6L 4L 2 - Dynamic problems MÜ + R U + KU F harmonic load F F e iωt ( K +iωr ω 2 M ) U F Frequency Response Function U - Properties of M and K Symmetric, (semi) definite, positive

One DOF example 5 x η m F k F 1 m 1 k 99, 11] η.2 Static problem kx F x F 1 k 11, 1 ] 9 Dynamic problem (Frequency response function) ( k(1 + iη) ω 2 m ) x F ] ] k ω 2 m ηk x r real problem: ηk k ω 2 m x i F ] k appears four times in the matrix! Direct interval resolution (Intlab) Analytic solution: x r x i ] (k ω 2 m)f (η 2 +1)k 2 2kω 2 m+ω 4 m 2 ηkf (η 2 +1)k 2 2kω 2 m+ω 4 m 2 x r x i x r x i ] ] ].3425,.3425].678,.396] overestimation ].24,.197].51,.397]

Formulation adapted to Finite Element Method 6 Young s modulus in E ] varying Stiffness matrix E k 11 k 12 k 21 k 22 ] ] Ek 11 Ek 12 E 1 k 11 E 2 k 12 Ek 21 Ek 22 E 3 k 21 E 4 k 22 E i E E i independent Symmetric General form: A] A ]+ N n1 ɛ na n ] {b {b + P p1 β p{b p K] K ]+ N n1 ɛ nk n ] FEM Definite (or semi-definite) Positive for ɛ n ɛ n K] corresponds to a stiffness matrix

Symmetry, Vertex 7 3.5 A] ] 3, 4] 1, 2] 1, 2] 3, 4] 3 2.5 x 2 2 P 4 P 1 ] 3 1 {x 2 4 { 1.5 1 1.5 P 3 1 P 2.5.5 1 1.5 2 2.5 3 3.5 x 1 { ] ] 38 1 1 2 + e 1 {x 1 6 2 2 { x 1 x 2 { 1 2, 1 9 5, 13 3 ] 3 ] { 44 2 e 1 1 e 1 e 1 1 x 1 1 3 x 2 13 3 x 1 1 2 x 2 5 2 x 1 7 15 x 2 9 5 sup bound sup bound inf bound inf bound

The algorithm of Rump 8 Rump s inclusion Fix point theorem Initialization ɛ (, 1) Y : g, g] A]{x {b {x {x + {x {x G]{x + {g X : 1 ɛ, 1+ɛ] Y Y : G X g with G] I] R]A] {g R]({b A]{x ) No Y X In practice R] A 1 ] {x R]{b Yes x Y

Algorithm adapted to the mechanical formulation 9 Taking into a account the factorization example with one (centered) parameter α A] A ]+αa 1 ] R] mid(a]) 1 A ] 1 {x R]{b A ] 1 {b Convergence ρ( G ) < 1 partition of e: e i e i ρ( G(e i ) ) < 1 e i x i x i {x 1 G]{x + {g G] I] R]A] αa ] 1 A1] {g R]({b A]{x ) αa ] 1 A 1 ]{x G] ɛ 1 A ] 1 A 1 ] ρ( G] ) ρ(η, ω) k 1 k (1+η 2 ) ω 2 m +ηk 1 ω 2 m (1+η 2 )k 2 2k ω 2 m+ω 4 m 2 3 25 {x n {x n 1 +( 1) n+1 α n+1 ɛ n B] n+1 {x B] A ] 1 A 1 ] ρ( G] ) 2 15 1 5.5.4 η.3.2.1 9 9.5 1 ω 1.5 11

The proposed algorithm 7 G e i B i Initialization ɛ (, 1) Y : g, g] ρ( G ) < 1 ρ( e i B i ) < 1 Yes X : 1 ɛ, 1+ɛ] Y Y : G X g No Y X No el e, ] A l A +.5eA 1 er, e] A r A +.5eA 1 Yes x Y x inf(x i ) x sup(x i )

Example 11 EI L 3 ]{ 12 6L d 6L 4L 2 θ { F M EI 1.8151, 1.9664]e4 (1.898e4 ± 2%) F 12, 98] M 44.1, 45.9] L 1 θ F d M ]{ { 2.178, 2.36] 1.18, 1.8] d 12, 98] 1e5 1.18, 1.8].72,.787] θ 44.1, 45.9] ]{ { 12 6 d 12, 98] EI 6 4 θ 44.1, 45.9] Rotation θ.5 1 x 1 3 verifylss Σ (A], {b) { d θ { d θ { 1.163,.44] e 3 1.282,.24] {.659,.494] e 3.381,.157].5 1 Hull of the mechanical set Mechanical exact solution set Σ (A], {b) 1.5 12 1 8 6 4 2 2 Displacement d x 1 4

Algorithm adapted to the mechanical formulation 12 x 1 4 Rotation θ.5 1 1.5 hull for the modified Rump s algorithm Solution sets for a, ] and,a] 1 x 1 3 2 2.5.5 verifylss 3 Rotation θ.5 Σ (A], {b) 3.5 4 4.5 Hull containing the mechanical solution 5 Exact mechanical solution exact hull Exact mechanical solution 7 6.5 6 5.5 5 4.5 Displacement d x 1 4 1 Σ (A], {b) 1.5 12 1 8 6 4 2 2 Displacement d x 1 4

Clamped free Bar. Comparison 13 x.5 x.75 x 1 E 1 E 2 x E 2 21 ± 1% GP a F Original Rump s algorithm Adapted algorithm 1.8 x 1 3 1.4 x 1 3 1.6 1.2 1.4 1 1.2 D 1 D.8.8.6.6.4.4.2.2.1.2.3.4.5.6.7.8.9 1 x.1.2.3.4.5.6.7.8.9 1 x

Example 1 14 f 1 f 2 k 1 k 2 k 3 m 1 m 2 Static Case ( ] ] k1 + k2 k2 k1 1 k2 k2 + + e 1 + k 3 ] ]) { k2 1 k2 1 e 2 + e k2 1 k2 1 3 k3 1 x 1 x 2 { Monte Carlo{ 1.1374,.11333] x MC1.18436,.19962] Monte Carlo{ 1.1399,.1136] x MC2.18454,.19925] Proposed algorithm {.1331,.11353] x lss.18394,.19968] x MC2 x MC1 real bound x lss f 1 f 2 x 1 x 2 Real part(m/n) Imaginary part (m/n).2.1.1.2 8.5 9 9.5 1 1.5 11 11.5 12 12.5 ω (rad/s).1.2.3 Dynamic Case ( ] (1 + iη 1 )k1 +(1+iη 2 )k2 (1 + iη 2 )k2 (1 + iη 2 )k2 (1 + iη 2 )k2 +(1+iη 3 )k3 ] ] (1 + iη 1 )k1 1 (1 + iη 2 )k2 1 (1 + iη 2 )k2 1 +e 1 +e 2 (1 + iη 2 )k2 1 (1 + iη 2 )k2 1 ] ]) { + e 3 ω 2 m 1 (1+iη 3 )k3 1 m 2 H 1 H 2 {.4 8.5 9 9.5 1 1.5 11 11.5 12 12.5 f 1 f 2 ω (rad/s)

Example 1 15.2 Real part (m/n).1.1.2 8.5 9 9.5 1 1.5 11 11.5 12 12.5 ω (rad/s) Imaginary part (m/n).1.2.3.4 8.5 9 9.5 1 1.5 11 11.5 12 12.5 ω (rad/s)

Example 2 : Truss structure 16 6 7 8 9 1 11 L1 L1 1 2 3 4 5 F 12 L2 L2 L2 L2 1 3 L1 L2 1m F 1 3 Nm I π 1 8 /4 m 4 S π 1 4 m 2 H(3, 3) (m/n) 1 4 1 5 E 21 ± 1% GP a 1 6 2 4 6 8 1 12 14 16 18 ω (rad/s)

Modulus of the FRF H(3, 3) of the truss. The stiffness is uncertain (E 21 ± 1% GPa). Min and max values, FRF for several values of E 17 1 3 1 4 H(3, 3) (m/n) 1 5 1 6 2 4 6 8 1 12 14 16 18 ω (rad/s)

Example 3: Multiple eigenvalues system 18 2 1 1 6 3 1 7 4 7 5 L 1m S π1 4 m 2 6 FRF H(1, 3) 1 8 1 9 ρ 78 kg/m 3 E 21 GP a 1 1 η 2% 1 11 For the first blade E E ± 1%.5 1 1.5 2 2.5 frequency ω (rad/s) x 1 5

Modulus of the transfer function H(1, 3) E E ± 1% for the first blade Crisp cases, envelope calculated with the modified algorithm 19 1 2 1 3 1 4 FRFH(1, 3) 1 5 1 6 1 7 5 1 15 2 25 frequency ω (rad/s)

Gyroscopic System 2 ]{ (J 1 ]{ Ψ + J B ) + J 1 Θ 1 ]{ Ψ aω + K z L 1 Θ 2 µ Ψ 1 Θ { r J B L Ω2 cos Ωt sin Ωt Ω m K y 1 x 1 5 1 2 K z 3 4 95 1 15 11 115 12 125 unbalanced rotating disk Ωandµ K y K z intervals 2 x 1 5 1 1 2 95 1 15 11 115 12 125 ( ] Ω M + K Ω G ηk Ω r M + e 1 Ω G + ηk Ω M + K Ω r G ] ]) { { Ω r G K 1 ηk 1 Ψ + e 2 Ω Ω r M ηk 1 K 1 Θ F r F i + e 1 Ω r { F r F i

Conclusion 21 Use of Interval Arithmetic overestimations resolution of linear systems Mechanical Formulation A] A ]+αa 1 ] Modification of the algorithm of Rump smallest Hull convergence efficiency robust envelopes Static cases Dynamic cases (FRF)