EP1991853A1

EP1991853A1 - Method for characterizing the endurance limit of a part from its surface profile

Info

Publication number: EP1991853A1
Application number: EP07731094A
Authority: EP
Inventors: Vivian Vergnes; Rémy CHEIRAGATTI; Catherine Mabru; Christine Espinosa; Monchai Surarachai
Original assignee: Airbus Operations SAS
Current assignee: Airbus Operations SAS
Priority date: 2006-03-07
Filing date: 2007-03-06
Publication date: 2008-11-19
Also published as: CA2645216A1; FR2898410A1; RU2467306C2; US8082114B2; JP5149204B2; CN101395461A; US20090093976A1; CA2645216C; WO2007101939A1; FR2898410B1; BRPI0707080A2; RU2008139621A; JP2009529135A

Abstract

The invention concerns a method for characterizing the endurance limit of a part from the state of its surface including the following steps: reading geometrical data describing the surface profile of the zone the endurance limit of which is to be determined, applying said data to a computing model so as to work out an estimate of the field of mechanical stresses in said zone of said part, deducing from said estimate of thee field of stresses at least one quantity characteristic of the endurance behaviour of the part.

Description

Method for characterizing the fatigue strength of a part from its surface profile

Field of the invention

The invention relates to the characterization of the fatigue strength of a part from its surface state.

Fatigue damage is a practical problem encountered in all types of parts subjected to a cyclic loading spectrum. However, the manufacturing processes of such parts cause fluctuations in the surface properties and therefore in the behavior of the parts in fatigue. Indeed, it is known that the fatigue resistance of mechanical parts depends in particular on their surface, where fatigue primers can appear. Most studies on the influence of the surface condition of a part on its fatigue behavior retain 3 parameters that can characterize it:

a geometrical parameter,

a metallurgical parameter, and / or

a mechanical parameter. Each parameter acts differently according to the material, so that it is generally possible to retain one as being the most representative for a given material.

In some cases, for example the aluminum alloy for aeronautical application mainly referred to here, the most representative criterion is the geometric criterion. This criterion is related to the shape of the surface roughness profile generated by the machining range. It is then common practice to apply a mechanical model to the geometric parameters thus identified in order to estimate their influence on the fatigue behavior. State of the art

Among the studies on the influence of surface states on fatigue behavior, those devoted to the influence of geometry use geometric models defining roughness parameters. These are intermediate means of description of the surface; indeed, mechanical models start from this geometric description, by these roughness parameters, to estimate its influence on the fatigue behavior. The diagram in Figure 1 schematizes this approach.

1) Geometric model:

The roughness parameters are calculated by the geometrical model, from a surface profile according to physical criteria or statistical criteria (see international standard ISO 4287 / 1-1984 (E / F / R) (1984)). This step, in general, is performed by a rugosimetry machine. Among the roughness parameters, those most often used to then determine the fatigue strength of the parts are:

- The average arithmetic difference (Ra) which is the area between the roughness profile and its mean line, or the integral of the absolute value of the roughness profile size above the evaluation length:

- The total deviation (Rt) which is a vertical deviation between the highest point and the lowest point of the roughness profile, over the total evaluation length.

- The average roughness depth (Rz) which is the arithmetic mean value of the simple depths Rzi of the consecutive sampling lengths (the symbol "i" designating the consecutive lengths that can be identified as having peaks in the surface profile - see Figure 2). 21 Mechanical model:

Roughness parameters thus obtained are then used by various known models to determine the mechanical properties. We can divide these models into 2 categories:

- models of the surface factor (Ks), and

- models of influence of roughness on fatigue behavior

2.1) Modelizations of the factor Ks:

The factor Ks is defined as the ratio of the endurance limit of the specimen given with a certain surface roughness and the endurance limit of a test specimen whose surface state is chosen as a reference.

K ₅ = ^ - ^σ D o &. Fatigue limit of the specimen whose surface state is chosen as a reference.

ODS: fatigue limit of the specimen given with a certain surface roughness. Several models have been proposed to define the value of Ks, called "Surface State Factor".

Stieler (1954), using the theory of stress concentration on geometric defects, proposed a formula of the type

or

C: is a factor dependent on the machining.

R: is defined as being equal to 2 Rt / Sg where Sg: represents the maximum thickness of materials involved in the process of priming a fatigue crack. Stieler has shown that it is of the order of magnitude of the grain size of materials in a rotary bending test.

Niemann and Glaubitz (1952) modeled their experimental results obtained in plane flexion by formulas of the type:

<j _D {Rts) where

Rts: is the roughness of the specimen given with a certain surface roughness

Rt: is the roughness of the reference specimen, σD: is the endurance limit of the reference specimen, σDS: is the endurance limit of the test specimen considered, n: is a coefficient depending on the material.

Brand et al (CETIM, 1980) constructed an abacus by smoothing a large number of available data, with straight lines of negative slope giving Ks (the surface state factor) as a function of the breaking strength Rm, for various values of roughness criterion Rt.

2.2) Modeling of the influence of the roughness on the resistance in fatigue:

Two approaches are generally used for the prediction of fatigue life as a function of roughness: an approach based on the notch effect,

- an approach based on the mechanics of fracture.

2.2.1) Notch effect

The notch effect approach uses the classical definition of the effective stress concentration factor, Kf, which is the ratio of the endurance limit σD of a smooth specimen to the endurance limit σD of the notched test piece. This coefficient Kf is generally lower than the coefficient Theoretical studies of stress concentration Kt. Some authors, notably Neuber, 1957, Smith et al, 1970, proposed relations between Kf and factor Kt. For low Kt values, Peterson (1959) established an empirical relation defining the factor Kf as follows:

where Kt: is the stress concentration factor,

α: is a constant related to the material p: is the notch radius.

Arola and Williams (2002) expressed Kt as a function of the roughness parameters Ra, Rt and Rz and the mean radius p with a notch bottom, and a parameter n which is a factor that depends on the type of loading (n = 1 for shear loads, and n = 2 for uniform tensile loads).

This factor Kt is then used to establish an empirical relationship defining the factor Kf.

2.2.2) Mechanics of rupture

Considering that the roughness of the surface forms nicks that can be likened to cracks, it is possible to use the results of the fracture mechanics. The fatigue behavior is then characterized by ΔKth which is the variation of the threshold of the stress intensity factor. Kitagawa (1976) then represented the evolution of this threshold ΔKth as a function of the length of the crack in a bi-logarithmic diagram on which the reference fatigue limit (on a polished specimen) is distinguished by a horizontal line, then a threshold curve that appears as a slope line -1/2. The transition between these lines defines an area where the experimental data deviates from the theoretical curves. Taylor and Clancy (1991) compared these curves to the roughness criterion Rmax, the predictions made by these two approaches to the experimental results. They concluded that for weak roughness, the fracture mechanics approach is suitable. While for high roughness, the notch effect approach gives better results. The value of Rmax corresponding to the intersection of these two theoretical curves can provide a validity limit of the prediction based on the fracture mechanics, which becomes too conservative beyond that. In all cases, Rmax seems to be the most significant surface criterion. Indeed, this criterion satisfactorily represents either the depth of the largest notch (the notch effect), or the largest length of the crack (mechanical failure).

Andrews and Sehitoglu (2000) focused on crack propagation and stress concentration factors present, distinguishing between cracks being considered short and long. They then proposed an expression of the stress concentrations taking into account a relaxation of the stress concentrations when the cuts are side by side.

2.3) Comments

In practice, the fatigue strength of the parts can be influenced, inter alia, by the geometric profile of the surface. Accidents in the shape of this profile influence the initiation or propagation of fatigue cracks. However, the models that take them into account do not start from the real form of these accidents but from simplified geometrical descriptions of the profilometric measurements. The parameters resulting from these descriptions are numerous, but none of them makes it possible to ensure for all the types of accident a relevance of the mechanical model which uses it. Sometimes it is wise to use one now and another, and only experience can decide a posteriori. This does not allow a characterization of the fatigue resistance without having carried out prior tests.

The subject of the invention is a new procedure for characterizing the fatigue strength of a part as a function of its surface state, allowing to get rid of any purely geometric description of this profile (by the classical parameters such as roughness coefficients Ra, Rt, Rz, etc.) to focus on a mechanical description of the part in more direct relationship with the fatigue behavior of the part under consideration.

Description of the invention

According to one aspect of the invention we start from a digitization of the 2D or 3D surface profile obtained by the current rugosimetry machines, to calculate directly modifications of the field of local stresses generated on the surface by this profile. It is therefore a matter of setting up a measurement chain of a mechanical criterion associated with the geometry of a surface to qualify it as fatigue. The invention thus proposes a method of characterizing the fatigue strength of a part from its surface state, comprising the following steps:

geometric data describing the surface profile of the zone whose fatigue resistance must be determined. * we apply these data to a calculation model so as to develop an estimate of the field of mechanical stresses in said area of said room,

from this estimation of the stress field at least one characteristic quantity of the fatigue behavior of the part is deduced. It should be noted that, since there is a direct estimation of the stress field from the acquired surface profile, without going through the determination of geometric coefficients, such as roughness coefficients, it is possible to arrive at an estimation of the behavior , considering that any accidents will not be neutralized by any geometric model. According to advantageous features of the invention, possibly combined:

the step of determining (or raising) the data characterizing the surface profile of the zone comprises a substep of measurement of the geometric profile of this zone, in practice by any known probe device; however, it may be envisaged to use other techniques, especially purely optical, electrical, sound, thermal,

the step of raising / determining the data characterizing this profile comprises a substep of sampling, which makes it possible to reduce the size of the storage memory required, without, however, risking completely neutralizing surface accidents; preferably, this substep of sampling is designed so as to reduce the number of data characterizing the surface profile by at least a factor of 10; however, the absence of sampling is possible if the calculation means allow it,

the step of determining the data characterizing this profile comprises a sub-step of filtering, for example to remove the effect of the inclination or the geometry of the part (for example for a cylindrical part, the curvature related to its diameter).

the step of determining the data characterizing this profile comprises a substep of adjustment according to the calculation model, which can guarantee the respect of the format imposed by the calculation model,

the calculation model to which the data are applied is a finite element calculation model or its variants (XFEM, BARSOOM, ....), which corresponds to a well-controlled calculation tool; other models of calculation can however be envisaged, such as in particular the other numerical methods like the particulate numerical models, the finite or spectral differences, the integral methods, - the model of computation integrates a thickness of the part; advantageously, this thickness is at least 0.5 mm, preferably at least 1 mm below the surface of said zone, which appears to be quite sufficient to correctly estimate the stress field, whatever the surface profiles studied; this thickness can nevertheless be optimized case by case according to the state of the art,

the calculation model determines, for each calculation element (finite element, or numerical element of the model used, etc.) stress values along two or three main axes of said zone, which allows a better estimate of the stress field than with a single axis,

the model is applied only at least a non-zero distance from the edges of the zone of the part, for example at least 1 mm from these edges, so as to avoid the effects of edges (unless it is possible to integrate these effects in the calculation model),

the characteristic magnitude of the fatigue behavior is a concentration coefficient of the maximum stresses, which corresponds to what is given by the current measurement chains using both a geometrical model and a mechanical model; other quantities such as the distribution of stresses in the thickness can be easily obtained from the numerical model used in the invention.

Objects, characteristics and advantages of the invention will become apparent from the description which follows, given by way of nonlimiting illustrative example, given with reference to the appended drawings in which:

FIG. 1 is a diagram schematizing the stages of the characterization of the fatigue strength of a part from its surface profile,

FIG. 2 is a diagram showing the roughness coefficient denoted Rz,

FIG. 3 is a diagram schematizing the steps of the method according to the invention, by analogy with the formalism used in FIG.

FIG. 4 is an implementation diagram of one embodiment of the method of the invention,

FIG. 5 is a diagram of a first step of implementing this method,

FIG. 6 is a graph showing the profile of a machined specimen over a length of 17.5 mm; FIG. 7 is a diagram of a second step of implementing the method;

FIG. 8 is a graph showing the acquired profile as well as the modified profile, FIG. 9 is a diagram of a third stage of implementation of the method,

FIG. 10 is a graph showing the modified profile as well as a detail thereof; FIG. 11 is a graph showing this modified profile, as well as this detail, broken down into finite elements;

FIG. 12 is a diagram of a fourth implementation step,

FIG. 13 is a graph representing the local stress field,

FIG. 14 is a diagram of a fifth stage of implementation of the method,

FIG. 15 is a graph showing the Wohler curve obtained for various specimens, and FIG. 16 is a graph showing the Wohler curve obtained for the same specimens after correction by the coefficient of stress determined by the coefficient obtained at the end of the fifth step.

According to the invention, a profilometric measurement is used in a mechanical model, without going through an intermediate geometric modeling of the profile involving the determination of roughness coefficients. This measurement chain is shown diagrammatically in FIG. 3: the mechanical model directly uses the profiling of the profile in a mechanical model, so as to determine one or more mechanical parameters, such as stress concentration coefficients.

The approach of this measurement chain is represented in FIG. 4, with the following steps:

step of acquiring the profile of the surface, by determining (or raising) data characterizing the surface profile of a chosen zone of the part under consideration,

- possible stage of treatment of the profile (it can, in some cases, be integrated in the data entry of the surface profile), integration of the profile possibly processed in a calculation model, which amounts to applying the data to a calculation model,

implementing the calculation model, so as to estimate the mechanical stress field in the area whose surface profile has been entered,

processing of the results of the calculation so as to deduce from this estimation of the stress field at least one characteristic quantity of this field (for example Kt) and therefore of the fatigue behavior of the part under consideration.

The details of these various steps can be summarized as follows.

Step 1: Acquire the Surface Profile The geometrical state of the surface is measured in this step (see Figure 5). It is determined by a measuring machine (contact or optical) of any appropriate known type, such as a probe device conforming to the standard NF-ISO 3274, June 1977, implementing the provisions of ISO - 4287 / 1984 (E / FR) 1984 already mentioned above. The resulting profile, ie the total profile (direct or gross profile) or the primary profile (in practice, after elimination of the nominal shape of the analyzed part area, and possible application of a filter low pass) is digitally recorded and is then used in the proposed measurement chain. FIG. 6 shows, by way of example, the profile of the surface of a machined specimen acquired by a rugosimetry machine of the "Mahr Perthometer-PKG 120" type.

An evaluation length of 17.50 mm was chosen as the reference length for all the surface ranges in the example considered here. The direct profile obtained without filter (the total profile, or direct) was recorded in ASCII format by an available function proposed by this roughness meter. This profile is composed of the form deviation, the undulation, the periodic or pseudoperiodic roughness (the striations and furrows) and the roughness aperiodic (tearing, tool marks and slits, tapping, etc.).

Step 2: Process the profile

The profile obtained in the first step is then advantageously sampled, modified and adjusted (see Figure 7). Sampling has the advantage of reducing the requested memory size and calculation time. Several sampling methods are possible, for example, with a fixed frequency, with an average neighborhood value, and so on. The profile can be modified by different filters to remove unwanted parameters such as tilt. A possible adjustment of the result may be necessary to be able to integrate this profile into the subsequent calculation model.

FIG. 8 represents, on a much finer scale than in FIG. 6, a detail of the profile acquired during the first step, as well as this same profile after treatment. In the example considered here, the processing step is performed by a calculation software. The sampling method is performed at a fixed frequency chosen to reduce the number of points on the order of 11000 points to about 550 points, no other modification being made. Then, the heights of the profile were adjusted by removing the average of the profile and the arithmetic difference of the profile. The first point and / or the last point were imposed by the respect of the condition to have a height equal to zero, with a step equivalent to the sampled frequency.

The software used in this step was the SCILAB software. Command lines are produced, and presented in Appendix A. It can be noted in the examination of Figure 8 that the profile obtained is smoothed, and some periodic or pseudoperiodic roughness and aperiodic roughness disappeared due to sampling.

Step 3: Integrate the profile into a calculation model, here a finite element calculation model

The modified profile is then integrated into a computational model for determining the stress field (see Figure 9). It is advantageously a finite element calculation model. Integration is different depending on the software chosen. Geometry, boundary conditions and assumptions are appropriately set so that the stress field can be calculated.

The transition from the processed profile (FIG. 10) to the finite element decomposition (FIG. 11) was performed using the SAMCEF-Asef calculation software. It needs an input databank with a specific format generated by a text editor.

In the example considered here, a thickness of 5 mm was taken into consideration by imposing a plane of symmetry, which corresponds to an equivalent thickness of 10 mm in the model (on either side of the line according to which the profile has been acquired). The calculation has been simplified by assuming linear elastic behavior and boundary conditions representing uniformly distributed loads. The sizes and the number of meshes were limited by the available memory area and the requested computation time. To facilitate the work in this step, command lines of the SCILAB software were made to generate the calculation file (databank) from the processed profile; they are presented in Appendix B.

Step 4: Calculation of the stress field, here by finite elements

This step (see Figure 12) is performed by finite element calculation software. One or more parameters are envisaged to represent the stress field thus obtained.

FIG. 13 represents the result obtained by the implementation of the finite element calculation software SAMCEF module asef chosen (see step 3) to calculate the field of the constraints. Only the group of results between the distance 1 mm and the distance 16.5 mm was retained, to eliminate the effect of the edge (on the sections 0-1 mm and 16.5-17.5 mm), with a depth of 1 mm. They have been registered for processing in the next step. Two results (the constraints in the main axes and the coordinates of the finite element considered) were recorded with post-processing commands ("post-process") in the database. Step 5: Process the results

Various processing can be applied here to the data from the computation step of the stress field, in order to deduce from the estimation of the stress field at least one characteristic quantity of the fatigue behavior of the part under consideration (less than explored area), such as stress concentration coefficients.

By way of example, the constraints of the field estimated in step 4 are divided by the nominal stress, so as to calculate the conventional stress concentration coefficients. One or more characteristic parameters of the influence of the geometrical state on the fatigue strength can then be calculated from these coefficients.

For example, the concentration coefficients of the maximum stresses are calculated.

An example of use of the results obtained is presented in FIGS. 15 and 16: results of fatigue tests, represented by the Wohler curve of FIG. 15, have been corrected by the concentration factors Kt of the maximum stresses obtained (see FIG. Figure 16). It is observed, by comparison of these figures 15 and 16, that the differences between the different surface states are largely attenuated by the correction (multiplication) by the measured Kt.

It is thus established that the proposed measurement chain is capable of providing quality indicators of the surface states of the parts intended to be subjected to mechanical loading. This measurement chain has the advantage of not passing through geometrical parameters and therefore does not require prior knowledge of the influence of a particular type of accident on the lifetime of the part. Appendix A: The Scilab Command Lines for Performing Step 2 Work

// Scilab

// Step 2 - Sampling and Edit Profile

// from a profile measured by Mahr-PGKl 20 (rugosimetry)

// the 01 June 2005 clear rugofile = 'profil_direct.TXT'; sample≈550; // Number of approximate sampling of the profile

[B, e] = mopen (rugofile, 'r'); if (er == O) then

// Import the mprintf profile file ('Import ^* % s ^s \ n', rugofile) rugo = tlist (['Point', mfscanf (en, '% s% s% s'))));

else dir≈ l; 1 = 0; while (r <>','),

I = l * 10 + str2code (r); r = mfscanf (fr, '% c'); end, end, resi = 0; pwr = 1; r = mfscanf (fr, '% c'); while (r <> code2str (-40)) & (r <> code2str (l 10)), res = resi + str2code (r) / (10 ** pwr); pwr = pwr + l; r = mfscanf (fr, '% c'); end, rugo (4) (ind) = dir * (l (1) + resi); err = MEOF (fr); end; mclose (fr); // Sampling sample = round (ind / sample); for i = l: ind if modulo (i, sample) == 0 then profile (j, l) = j + l; profiiα _I 2) = rugo (2) (i); profilo, 3) = rugo (4) (i); j = j + i; end, end;

// change profile average = mean (profile (:, 3)); arith = mad (profile (:, 3)); profile (:, 3) = profile (:, 3) - (mean + arith); profflG, l) = j + l; promG, 2) = profiiα-l, 2) + (profilo-l, 2) -profile (j-2.2)); proFil (i, 3) = 0; save ( 'profil.dat'profile);end;

Appendix B: The Scilab Command Lines for Performing Step 3 Work

// Scilab

// Step 3 - Generate Finite Element Calculation Bankfile (Samcef-asef)

// Ie Ol June 2005 clear bankfile = 'essai.dat'; dquote = ascii (34); fd = mopen ('bankfile,' w '); load ( 'profiLdat');

// Generate Bank File for calculation by Samcef (Asef)

// Prerequisite mfbrintfffd '\ ι ******************************************* ****** \not

\! * Model EFM-2D * \ n ... \! * Obj: Calculate the concentration of the constraint * \ n ...

\! * Material: Aluminum-elastic * \ n ...

\! * Model: Measured area * \ n ...

M * Hypothesis: Flat deformation * \ n ...

\! * Mesh: Direct tranfîni * \ n ...

W * CL: traction & traction each end * \ n ... yi ************************************ ************** _n

M * Author: * \ n ...

\! * the% s * \ n ... χ _\ ************************************ ************** _n yι ********************************** *************** Y _n I ₍ Ja _W ))

// Variable mfprintf variable (fd, '. Del. * \ N ...

\not...

\! \not...

\! Abbreviation for \ n ...

\! =====. ======= - ============== V _n ...

\! \not...

\! General geometry \ n ...

\ Ι_. __ \ _n

ABRE Λ \ thick ^{Λ Λ} 5000 ^λ \ t \! Thickness \ (um \) \ n ...

ABRE Λ \ sec2 ^{^} '1000' \ t \! Smooth surface (um) \ n ...

ABRE Λ \ mf_y '' 500 '\ t \! Thickness of area fine mesh (um) \ n ...

ABRE 'WIm' '% 0.2f \ t \! Profile length \ n ...

\! \not...

\! Feature of the Material \ n ...

\ n \ n

ABRE '\\ E_alu' '70000' \! Aluminum Young modulas \ (N / mm; MPa \) \ n., ABRE '\\ poi_alu' '0.33' \! Coefficient of poisons \ n ...

\! \not...

\! Number of items \ n ...

\ Ι. _. U ₁

ABRE '\\ NE_surf' 500 '\ t \! On the profile \ n ... ABRE '\\ NE_sec2''(Wsec2: (\\ lm: \\ NEjsurf))' \ t \! Smooth surface \ n ... ABRE '\\ NE_fmY''(\\ mf_y: (\\ lm: \\ NE_surf))' \ t \! Fine mesh area \ n ... ABRE '\\ NE_groY' ((\\ thick - \\ mf_y): \ (\\ lm: \\ NE_surf)) '\ n'_; profil0,2)) // Geometry mfprintf (fd _; '\! \ N ...

\! A. Geometry \ n ...

\! Vt ₁ ...

\! A.l.2D-geometry \ n ...

\! \not...

.Point \ n ...

\! Coordinates of the profile \ n ') mfprintf (fd, V) mφrintf (fd,' I% 4i \ t X% 9.2f \ t Y% 9.4f Vn ', 1,0,0) mφrintf (fd,' I% 4i \ t X% 9.2f \ t Y% 9.4f Vn ', profile) mφrintf (fd,' V! \ n ...

.Spline Vn ...

11 Point 1 a% i \ n ...

Vn ...

.Point \ n ... i 8001 x 0 y- (Wmf_y) Vn ... i 8002Rx 0 y - (Wepais) Vn ... i 8003 x - (\ Ysec2) yθ \ n ... i 8004Rx 0 y - (Wmf_y) Vn ... i 8005Rx 0 y - (Wepais) Vn ... i 8007x (WIm) y - (\ Vmf_y) Vn ... i 8008Rx 0 y - (Wepais) \ n ... i 8009 x (Wlm + Wsec2) yθ \ n ... i 8010Rx 0 y- (Wmf_y) Vn ... i δOURx O y - (Wepais) Go ...

Vn ... .DROITVa ... i 103 Item 180018002 Vn ... i 105 Item% i 80078008 Va ... i 107 Item 8003 to 8005 Vn ... i 109 Item 8009 to 801 IVn ... i 111 Number 8003 IVn ... i 112 Point% i 8009 Vn ... i 113 Point 8005800280088011 Vn ... V! Loading line Vn ... i 121 Item 80038005 V! On the left Vn ... i 122 Point 80098011 \! On the right Vn ... i 123 Point 80058011 V! Below Vn'j + 1 d + 1 d + 1)

// Mesh mφrintf (fd, 'V! Vn ... V! Vn ... V! A.2. Mesh Vn ...

V! Vn ...

.CONTOURVn ... i 1 Line 1051061141041031 Vn ... i 2 Line 103104113108107111 Vn ... i3 Line 105106115110109112 Vn ... Vn ...

.Domain Auto \ n ... Vn ... .GEN Vi ...

VMwidth in horizontal (X axis) \ n ... modifies Line 111 113 \ t Element (\\ NE_sec2: 3) distribute 3 6 Vn ... modifies Line 112 115 \ t Element (\\ NE_sec2: 3) distribute 2 6 Vn ... modifies Line 1 \ t Element (\\ NE_surf) Vn ... modifies Line 114 \ t Element (\\ NE_surf) \ n ... VIMalize in vertical (Y axis) \ n ... modifies Line 107 103 105 109 \ t Element (\\ NE_finY) \ n ... modify Line 108 104 106 110 \ t Element (\\ NE_groY: 3) distribute 2 6 \ n .. \ n ...

VIMautomatic \ n ... deg 1 \ n ... cl 1 c2105106 c3114 c4104103 \ n ... mesh 1 transfinite \ n ... cl 111 c2103104 c3113 c4108107 \ n ... mesh 2 transfinite \ n ... cl 112 c2 109 110 c3 115 c4 106 105 \ n ... mesh 3 transfinite \ n ... \ n ... \! Vn ...

V! A3. Mesh changes \ n ...

\ _n

Vn ')

// Hypothesis and Selection Group mfprintf (fd, \\ Vn ...

\! B. Hypothesis and Material \ n ...

\! ^~ - = = = = \ n ...

Vn ... .MAT Vn ...

1 1 Name 'Alu_elasπV \ n ...

Beha% cElastic% c Vn ...

Yt (\\ E_Alu) \ n ...

Nt (\\ Poi_Alu) Vn ... \ n ... .AEL Vn ...

ATTRIBUTE 1 to 3 MAT 1 \ n ... \ n ...

.hyp DEFO PLAN \ n ... \ n ...

MSelection of groups \ n ... .SEL \ n ... \ n ...

Group 1 nodes name% cResultat_zone% c \ n ... STRUCTURE BOX $ \ n ... XI (1000) XS (Wlm-1000) $ Vn ... YI (100) YS (-1000) $ \ n .. ZI - (l) ZS (l) \ n ...

Vn 'dquote, dquote, dquote, dquote) // Limit Conditions mfprin.tf (fdΛ! \ N ...

\! C. Limit conditions \ n ...

.CLM \ n ... \ n ... \! load case 1 'pull' \ n ... cha Line 122 con 100 compo 1 ne 1 \ n ...

Fix Line 121 compo 1 \ n ...

Fix Line 123 compo 2 4 6 \ n ... Vn ...

\! General data \ n ... \! \not...

.sam nop5 1 nop6 1 \ n ... .fin 1 \ n ')

// Post-processing mφrintf (fd \ ========= ⁼ ===== ==== ^{^≈: ~} - - ====== \ n ...!

-Post & \ n ...

.DeL * \ n ...

.doc db% cessai% c \ n ...

Assign FAC% cessai_as% c \ n ...

.des \ n ...

Disc -1 \ n ... trace mode =% cresult.txt% c \ n ... code 1411 comp 1 \ n ... group 2 \ n ... list \ n ... trace mode 0 \ n. .. trace mode =% cnoeuds.txt% c \ n ...

.noe load group 2 \ n ... list \ n ... trace mode 0 \ n ...

.stop \ n '...

, dquote, dquote, dquote, dquote, dquote, dquote, dquote, dquote) mclose (fd); mprintf ('Export'% s '\ n', bankfile)

Claims

1. A method of characterizing the fatigue strength of a workpiece from its surface state comprising the following steps:

geometric data describing the surface profile of the zone whose fatigue resistance must be determined,

* we apply these data to a calculation model so as to develop an estimate of the field of mechanical stresses in said area of said room,

from this estimation of the stress field at least one characteristic quantity of the fatigue behavior of the part is deduced.

2. Method according to claim 1, characterized in that the step of raising the data characterizing the surface profile of the zone comprises a sub-step of measuring the geometric profile of this zone.

3. Method according to claim 2, characterized in that the data recovery step characterizing this profile comprises a sub-sampling step.

4. Method according to claim 3, characterized in that the sampling substep is designed so as to reduce the number of data characterizing the surface profile by at least a factor of 10.

5. Method according to any one of claims 2 to 4, characterized in that the data recovery step characterizing this profile comprises a substep of filtering.

6. Method according to any one of claims 2 to 5, characterized in that the data recovery step characterizing this profile comprises a substep adjustment according to the calculation model.

7. Method according to any one of claims 1 to 6, characterized in that the calculation model to which the data are applied is a finite element calculation model.

8. Method according to claim 7, characterized in that the calculation model incorporates a thickness of the workpiece of at least 0.5 mm below the surface of said zone.

9. Process according to any one of claims 1 to 8, characterized in that the calculation model determines, for each calculation element, stress values according to at least two main axes of said zone.

10. Method according to any one of claims 1 to 9, characterized in that the model is applied only at least a non-zero distance from the edges of the part area.

11. Method according to any one of claims 1 to 10, characterized in that the characteristic magnitude of the fatigue behavior is the maximum stress concentration coefficient.