Systems Engineering
Sensitivity analysis
Structural reliability
In order to evaluate the risk of a technological solution and to
optimize costs in design or maintenance
2
General presentation
General presentation
Public company with 110 k€ of funds
Location : Clermont-Ferrand - France
Prize winner in 2001 and 2003 for its innovative developments
Company created in 2001
22 persons
Scientific team founder
Prof. Maurice LEMAIRE
Délégué scientifique
Engineer INSA Lyon
IFMA Research Director
Maurice PENDOLA
CEO
Doctor & Engineer IFMA
Reliability Expert
3 Paris Clermont-Ferrand Bordeaux Seyne / Mer Paris Clermont-Ferrand Bordeaux Seyne / Mer
Parc Technologique de la Pardieu 1 allée Alan Turing
F-63170 AUBIERE Tél (00 33) 4 73 28 93 66 Fax (00 33) 4 73 28 95 76 aubiè[email protected]
Subsidiary Technopôle Var Matin
Route de la Seyne F-83191 OLLIOULES Tél (00 33) 4 94 62 51 95 Fax (00 33) 4 94 62 59 47 [email protected]
Localization
! New Adress ! From January 03, 2008Centre d'affaires du Zenith 34 rue de Sarlièvre F-63200 COURNON d'Auvergne
4 4
Our vocation, Our ambition
« Expert in probability applied to engineering, PHIMECA aims to
become the international reference in this field»
Maurice PENDOLA, CEO PHIMECA
De grandes sociétés nous font déjà confiance:
Des références scientifiques nationales et internationales:
2 European projects
4 ANR projects
ImdR actor
More than 100 communications/articles
Some references
5
Ours Knowhows
6
Modelling & Engineering systems
From drafting to complete design, PHIMECA offers adapted solutions
Expertise:
CAD modelling
Structural analyses by finite element (elasticity, plasticity, fracture, fatigue, stability, thermal and dynamic)
Design according to codes of practices Integrity justification
Durability estimation Drafting
Tools:
CAD : SolidWorks 2007, CATIA V5, I-deas
Simulations : ANSYS V11, ABAQUS, NASTRAN NX Clusters
7
Dedicated Software solutions
Software development of high-tech tools
From simple Graphical User Interface (GUI) to high performance solvers, our computer science department research and develop yours tools.
8
General sketch for uncertainty analysis
9
Reliability by PHIMECA Engineering
Interest of this approach
Control of possible outcomes for a given design Prediction of success and safety margins
Taking into account of the randomness on the input parameters Quantification and hierarchization of input parameters (driving a probable failure)
10
PHIMECA Software
loading, time
Probability
of failure Failure probabilityReliability index
Second order approximations (SORM) Most Probable Failure Point (U et X) Parametric study
Safety factors Safety
factors Partial safety factors
Importance factors
Sensitivities and elasticities with respect to distribution parameters
11
Sensitivity analysis / Reliability analysis
Sensitivity analysis / Reliability analysis
Sensibility analysis =>
probability distribution of an output
What’s the shape of the distribution ? What’s the influence of input variables ?
Numerical methods:
- Response surfaces (polynomial chaos) - FORM, ...
Reliability analysis =>
probability to exceed a treeshold for a failure event
Fractile ?
Assess the level of component's reliability Characterization & ranking of the impact of uncertainties on the size of a component
12
Exhaust manifold
Objectives :
• quantification of the reliability;
• quantification of the parameters importance;
• distribution of the lifetime of the manifold.
Manufacturing processes
and
diversity of the customers
introduce great variability in
design parameter values and then influence the
failure risk
of the exhaust manifold
Develop a
complete probabilistic framework
in
fatigue design
of an exhaust manifold,
13
Application to the manifold (1/2)
Probabilistic parameters
•
Geometry
: 6 thicknesses overall the manifold
•
Material
: Young’s modulus
E
, yield strength
S
E, hardening coefficients
C ,D
(dependent from temperature)
•
Fatigue randomness
expressed by
ξ
(independent from temperature)
•
Loadings
: minimal
T
minand the maximal temperatures
T
maxThermo-mechanical (FE models)
simulation of the behavior
Probability of failure
P (G({X}) ≤ 0) = P (Nf (∆εp {X})− Nrequired ≤ 0)
Evaluation of ∆
ε
pat a
given node
Random fatigue model
(approximately 200000 nodes)
Thermo-mechanical (FE models)
simulation of the behavior
Probability of failure
P (G({X}) ≤ 0) = P (Nf (∆εp {X})− Nrequired ≤ 0)
Thermo-mechanical (FE models)
simulation of the behavior
Probability of failure
P (G({X}) ≤ 0) = P (Nf (∆εp {X})− Nrequired ≤ 0)
Evaluation of ∆
ε
pat a
given node
Random fatigue model
14
Application to the manifold (2/2)
Reliability results:
-0,023 0,025 0,075 -0,001 -0,002 0,011 0,007 -0,084 0,033 0,021 0,087 0,977 -0,153 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 1,2ep1 ep2 ep3 ep4 ep5 ep6 T max T ral uE uC uD uSE uNr
Direction cosines
• Failure probability: 2.3% (43 FE calls)
•
Most important parameters:
o Fatigue randomness
ξ
o
T
maxo Yield strength
S
Eo Young modulus
E
o Plastic coefficient
C
Reliability results:
-0,023 0,025 0,075 -0,001 -0,002 0,011 0,007 -0,084 0,033 0,021 0,087 0,977 -0,153 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 1,2ep1 ep2 ep3 ep4 ep5 ep6 T max T ral uE uC uD uSE uNr
Direction cosines
-0,023 0,025 0,075 -0,001 -0,002 0,011 0,007 -0,084 0,033 0,021 0,087 0,977 -0,153 -0,4 -0,2 0,0 0,2 0,4 0,6 0,8 1,0 1,2ep1 ep2 ep3 ep4 ep5 ep6 T max T ral uE uC uD uSE uNr