KR100324900B1 - Method for fatigue life prediction(FLP) - Google Patents

Abstract

The present invention relates to a fatigue life prediction method for predicting the life expectancy according to the type of specimen by using the material properties of each material to enable the fatigue life prediction and safety diagnosis of the industrial structure, in particular by using the material properties of each material In predicting lifespan, crack (itiation) crack initiation and crack growth analysis are combined.

The present invention integrates most of the existing crack growth models in crack growth analysis and allows most of the other models to be practically combined with each other to meet the respective analysis requirements. We have also included improved models with the concept of interactive zones, which the inventors have recently invented and which no other analysis program permits, to ensure that models and their parameters are used more consistently. For this purpose, the da / dN test data processing function (curve fitting) is included in an integrated package.

Description

Fatigue life prediction method {Method for fatigue life prediction (FLP)}

The present invention relates to a fatigue life prediction method for enabling fatigue life prediction and safety diagnosis of industrial structures. In particular, the prediction of crack initiation in the prediction of the lifespan according to the type of specimen using the properties of each material (fatigue (crack)) Fatigue life prediction method integrating initiation and crack growth analysis into one.

Fatigue is the most common form of failure of structures and one of the most difficult to handle. Material fatigue is a gradual physical process. A common feature of fatigue is that the structure functions for a considerable lifetime and thus is exposed to external activities repeatedly every day. Each period of activity is a small but irreversible cause of fatigue progression. Fatigue is apparently occurring even below the linear elastic limit and local breaking strength. The material can, however, break even under static loads called static strength, and fatigue is related to periodic repeated loads that are certainly less than static strength. The material may weaken under periodic loads and eventually break down even at loads lower than static strength. This indicates that the fatigue strength (fatigue limit or durability) of the material is completely different from the static strength of the material and is much lower than the static strength of the material.

During fatigue, damage accumulates in the material. At the microscopic level, this may be the accumulation of microscopic stresses in the surroundings, such as the accumulation of defects, interfaces or inclusions, or monolayers. In the macro view, the damage is a slip band or plastic propagation, microcracks or voids that together cause a pronounced macro crack. The crack then grows under fatigue loading until the final fracture.

Fatigue crack origins always begin with discontinuities caused by any reason due to load distribution, geometry of structural components or microstructure of heterogeneous materials. This reveals an important feature that fatigue is a very local phenomenon.

Although the understanding of fatigue progress has been greatly improved, fatigue remains a very common problem for engineers who need to design, build, operate and maintain components and structures.

There are many factors that affect fatigue life, including the characteristics of service loads, material properties, stress distributions in components, surface finish of components, working environment and damage accumulation processes. In the fatigue life analysis model, factors affecting life should be considered in several ways in life estimation.

Traditionally, small applied loads also result in small plastic strains, which can be ignored as a whole, and thus the number of cycles (fatigue life) leading to failure is large (typically> 100000). So this is classified as High Cycle Fatigue (hereinafter referred to as HCF). At higher loads, on the other hand, plastic deformation is comparable to elastic deformation, and parts break in less cycles, which is referred to as low cycle fatigue (hereinafter referred to as LCF). As such, HCF and LCF have different characteristics, but in actual applications with irregular loading conditions, it is sometimes difficult to distinguish them from each other.

The fatigue failure process in metal materials is divided into three stages: crack initiation, crack growth and final failure. Thus, the fatigue life of a component or structure usually consists of two parts: crack initiation life and crack growth (propagation) life. Although this has been generally accepted, the exact definition of what is the crack initiation is still controversial.

Engineers in the maintenance and overhaul areas in particular prefer to define the emergence of engineered cracks that can detect crack initiation. Fracture mechanics analysts, on the other hand, regard crack initiation as micro-nucleation.

Crack initiation life can usually be handled by traditional fatigue approaches. The place and time of initiation have some uncertainty due to inconsistent nature and local features of fatigue. Initiation life accounts for the majority (50-80%) of the total fatigue life of the structure according to the structure type and initiation definition.

Indeed, many studies of fatigue do not distinguish between crack initiation and propagation, and traditional fatigue approaches typically consider all of them as fatigue life. Only after failure mechanics were well established and used to analyze crack growth, traditional fatigue approaches began to be used for crack initiation analysis. Continuous damage mechanics have also been used in crack initiation analysis but are still not accepted by the engineering community.

The second stage of fatigue, called the stable crack growth stage, also consumes 20-80% of the total fatigue life, depending on the definition of initiation and the properties of the structural components.

On the other hand, the third stage of unstable breakdown leads to the final breakdown of the mechanical parts.

Although short crack behavior should also be carefully considered, Fracture Mechanics (FM) concepts can be used to characterize the growth of the crack and the characteristics of the resulting fractures.

There are two different approaches to fatigue life prediction: the traditional fatigue approach based on empirical life curves (Palmgren-Miner method, hereinafter referred to as PM approach) and empirical da / dN-. This is the fracture mechanics approach (called the FM approach) with a curve (Paris-Erdogan method). The PM approach can predict the total or starting life of an element depending on how the life curve is obtained. The FM approach, however, deals only with propagation phases.

Before discussing detailed fatigue life prediction methods, a general concept is outlined.

As a general consideration, for most empirical fatigue investigations, the specimen is usually tested under sinusoidal stress changes with constant amplitude, but the actual external load is always irregular. Therefore, in order to apply the experimental results of the laboratory to the prediction of fatigue failure in the structure, some assumptions about the damage contribution due to the continuous stress cycles with varying amplitudes are necessary.

In the FM approach, the damage d affected by each cycle can be easily defined as physical sound and measurable crack length increments da. However, in the traditional PM approach, physically significant damage has not yet been defined. Only generally accepted damage is assumed to be d = 1 / Ni. Where Ni is the number of cycles to failure determined by the appropriate Wohler curve or S-N plot

Damage to all (or half) cycles is assumed to comply with the law of linear accumulation (Miner's law). That is, the total damage is the sum of the damage by all cycles. This is the same for both the PM and FM approaches. The crack length in the FM approach is exactly the sum of all crack increments. However, the PM approach also follows this rule where there is no evidence to prove the estimated fatigue damage. There are other accumulation rules proposed, but they also lack a clear physical meaning. Therefore, Miner rules are still the most widely used because of their ease of interpretation.

Critical conditions are used to terminate life calculations. Critical damage conditions in the PM approach are considered as D = 1 based on the theoretically assumed damage definition. The change to D = constant, called the relative Miner law, is often done to account for changes in the actual load spectrum. In the FM approach, the critical condition can be fracture toughness (Kc) or critical crack length, both of which are meaningful material properties. The FM approach also has the flexibility to calculate (as needed) the lifetime to the conditional crack length.

Both the PM and FM approaches rely on laboratory tests to characterize material properties. However, the method of characterizing material properties is completely different. The PM approach is based directly on the life curve tested, while the fracture mechanics approach is based on crack growth data da / dN curves. Curve fitting techniques are commonly used to provide material properties in order to obtain some approximate shape solution. Details of these curves will be explained later when discussing different prediction methods.

First, the PM approach is described. There are nominal stress approaches and local strain methods.

The nominal stress (SN curve) approach shows that the structure is very small (nominal only), although the force exceeds the yield stress at the stress raiser (where stress concentrations occur in the specimen or structure). Mainly used for HCF subjected to elastic stress. This approach has been employed for a long time resulting in many S-N curves for different materials under different conditions.

The Wohler SN curve is obtained for a specific geometry (ie stress concentration factor Kt). This is the number of cycles at break (or at crack initiation) Indicates. Every cycle has a fixed stress range or amplitude in the test and one specimen represents one point on the curve. Thus, one SN curve requires several specimens. Since the SN curve is geometric dependent, different SN curves should be used for different stress concentrations. 1 shows a typical SN curve. Below the stress range S is longer than the life N. All curves are analytical, double-logarithmic and based on the basic curve I, expressed below.

Local deformation methods, on the other hand, are primarily considered for use in LCFs where the local force in the structure exceeds the yield stress. This method has been used successfully to predict component fatigue life. Although the local deformation method is similar in concept to the S-N trunk line method, it differs from the S-N curve method in several ways.

Strain-life curves are also generally obtained by experience, but the data are processed according to the strain component. Deformation can be divided into elastic and plastic parts, and the life curve is generally consistent with the Manson-Coffin equation below.

here, , , b and c represent material properties, fatigue strength and ductility coefficients and indexes, respectively.

The basic assumption behind the application of fracture mechanics to fatigue is that crack growth is related to the change in stress intensity factor K. The calculation of K is common to all FM methods.

In general, the stress intensity factor is expressed as

Where β is the geometry factor for a given crack shape.

To describe the superposition of different geometrical effects, β can be rewritten in the form

FM approaches include non-retardation methods and delay / acceleration methods. Non-delay methods include Paris, Forman, and Walker. Delay / acceleration methods include Wheeler, Willenborg, Elber's crack closure concept, Willenborg-Chang, and HeQingZhi.

First, the non-delay method is described.

In the non-delay method, the Paris-Erdogan law of crack growth (Paris method) is the most common method.

Where C and m are empirical constants derived from laboratory tests to specify the stress ratio R.

Since the Paris method does not allow changes in the stress ratio R, Forman proposed the following equation to consider both the effect of rapid crack growth close to the stress ratio R and the unstable state Kc.

The Walker method includes the negative stress ratio R part and treats them as follows.

Where C, m, p and q are the properties of the material derived from test data under different stress ratios, respectively.

Describe the delay / acceleration method.

Delay / acceleration models consider the interaction between different load cycles. So the fatigue life should be normal integrated every cycle.

In the Wheeler model, only delays caused by large overload cycles are considered by simply multiplying the initial crack growth rate by the delay coefficient, such as da / dN (ret.) = Cw ^* da / dN (no ret.). The retardation coefficient is calculated from the concept of the plastic zone size r in front of the crack tip.

The subscripts i and 0 here describe the parameters in the current cycle and in the overload cycle. The index n is determined by experience.

Elber paid attention to crack closure for the first time and proposed the effective stress range concept to consider the effect of stress ratio for the first time. The effective stress magnification range is expressed as follows.

Elber also expressed the following for an aluminum 2024-T3 sheet used as an aircraft material:

Newman further developed this model to calculate crack open stress for one cycle and one cycle through FEM for spectrum loading. The calculation of crack closure by one cycle of FEM cycles is too expensive and engineering complex, and some have proposed a constant crack closure model, but this is too simple to consider the actual interaction between individual cycles. The ideal model should be able to easily calculate the crack closure (opening) stress, but still calculate one cycle one cycle.

Willenborg's model was originally a modification of Wheeler's model by combining the size of the plastic zone with the concept of effective stress. The crack growth rate delayed due to the excess load is the effective stress ratio And effective stress magnification range Is corrected by substituting for the equation of Forman.

Wow The equation for calculating is

here, Silver plane stress, Is for plane deformation.

The generalized Willenborg model, modified from Equations 13 and 14, is as follows.

here, Is Threshold for, _c is the cut-off value for excess load, both of which are properties of the material determined by the test data.

Both the Wheeler and Willenborg models can only calculate the delay effects caused by positive overloads.

Chang modified the Willenborg model by using a Walker method with a negative effective stress ratio to account for the positive and negative R effects, which is commonly called the Willenborg-Chang model. All of the equations used are the same as the Willenborg model, except in Eq. 15 the negative values are left unchanged.

The HeQingZhi model uses the residual compress stress concept in the plastic zone and also takes into account stress relaxation. Any crack growth equation can be used with this model as long as the effective stress magnification range is calculated according to the HeQZ method.

The correction factor U for the effective stress extension range is formulated as:

Is the stress relaxation coefficient and is expressed as follows.

However, all computer analysis programs that predict fatigue life using the fatigue life prediction model of the structures described above treat the traditional fatigue (crack initiation) problem and the crack growth problem separately.

Also, when it comes to crack growth analysis programs, some analysis programs incorporate specific models but have little ability to integrate most existing models. Broek's program, for example, does not include Elber's crack closure model.

And most analysis programs do not include test data processing routines as an integral part, but allow users to manually process test data or use other software packages.

Therefore, there is a need for an integrated fatigue life prediction method that can integrate crack models and analyze crack initiation and crack growth simultaneously.

An object of the present invention developed by such a demand is to provide a fatigue life prediction method incorporating the prediction of crack initiation and the crack growth analysis in predicting the lifespan according to the specimen type using the properties of each material.

1 shows a typical S-N curve

2 is an overall flow chart of the fatigue life prediction method of the present invention

3 is a flow chart of fatigue onset prediction

4 is a flow chart of crack growth analysis

5 is a flowchart illustrating a life calculation process of FIG.

6 is a diagram showing a main screen of an application program embodying the fatigue life prediction method of the present invention;

7a and 7b is a comparison of the results of fatigue life using FLP and constant fatigue load at constant amplitude load

8a and 8b is a comparison of the results of the fatigue test using the FLP and the fatigue test under excessive / underload

In order to achieve the above object, the fatigue life prediction method of the present invention is integrated to analyze the crack initiation and growth analysis at the same time, and has the following characteristics.

First, when it comes to crack growth analysis, the fatigue life prediction method of the present invention integrates most existing models in crack growth analysis, and furthermore, it is possible to combine most different models practically together to suit different analysis requirements. It was.

Second, we included improved models with the concept of interactive zones that the inventors have not recently allowed in the recently developed Chen-Lee model.

Third, the fatigue life prediction method of the present invention includes da / dN test data processing (curve fitting) in an integrated package to ensure that the models and their parameters are used more consistently. Developed in station and Visual C ++ languages.

The fatigue life prediction method of the present invention is divided into two parts, a fatigue initiation prediction that predicts the first life of cracking in a defect-free structure and a crack growth analysis that predicts the life of crack growth, that is, a prediction of the life to fracture. have. In addition, it is divided into the type of method (RateFoumula), the state of stress (Correction), the type of specimen (Geometry), etc., and the threshold effect can be analyzed.

Closed models include Elber, Newman, Davidson, Chen-Lee, Elber-Chen-Lee, and others. Here, Chen-Lee et al. Models are not released yet.

The Chen-Lee model, etc., considers Elber's crack closure effect, which, briefly, contains the effects of plastic stress due to overload and residual stresses related to Elber's closure concept, which are important for predicting the effects of load interactions. In other words, the overload delay phenomena assuming that the crack closing force increases after overload. In addition, the fatigue crack growth life prediction of the existing models is relatively applicable to one overload / underload compared to the conventional retardation model, but at several overload / underload (more irregular) Contrary to poor fit, the model can be successfully applied to complex loading conditions.

The interaction model is also a prototype, refinement of commonly known models such as the Wheeler model and the Willenborg model, I.Z. It is classified to include concepts, etc., and it is possible to analyze.For the types of specimens, representative specimens specified by ASTM (American Society for Testing & Materials) -CCT (Center Cracked Tension), CT (Compact Tension), 3 points Specimens of various shapes such as bend (3 point bend) and surface cracks can be predicted.

The fatigue life prediction method of the present invention is composed of many modules, and each module performs different tasks, and thus has different input files and output files. Also controlled by the user's menu item selection and / or input data, the following tasks can be performed.

1. CCT, CT, SENB (3 point bend) of the standard specimen, styryl funds panel (stiffend panel) certain pre-run the geometric configurations, including - necessary data, for example, the width, thickness and initial crack in the fatigue test Calculate the shape functions used in the stress intensity factor (SIF) for length, final crack length, etc. as a file or directly through the input window, and calculate the undefined stress distribution along the cracked line. Calculate using the integral.

Specifically, select Crack Growth from the menu, then select Geometry from the submenu, enter the file name in response to the geometry-data-file name input request, and in the next step, Width, Initial Crack Length, and Final Crack. Entering length, thickness, etc. calculates the 'shape function' value for the crack length. The calculated value is stored as a geometry-data-file name with a .dat extension.

2. Process the constant amplitude (CA) fatigue crack growth test (aN) data to obtain the da / dN relationship for the SIF ranges for the geometries described above. In this case, numerical tables may be used.

In detail, select the crack growth in the menu, then select 'RateData' and 'from file' in the sub-menus sequentially, and in response to the input request of the rate-data-file name. If a file name is entered, the da / dN value is calculated for the SIF range. The calculated value is stored as a rate-data-file name with a .dat extension.

3. Fit the approaches of the closest form and determine the corresponding parameters. The rules of Paris are fixed, and you can use Paris, Forman, Walker, Forman-Walker or simply a check at your option.

In detail, select the crack growth from the menu item and then select 'Rateformula' from the submenu to select an existing method for performing parametric fitting.

If the basic model needs to be changed, i.e. if it is analyzed for irregular loads such as the actual load conditions, select 'correction' from the sub-menu and select the appropriate change model. Modifications include thresholds, stress states, crack closures, and various load interaction models (both delay and plasticity). If you need to make some other changes at the same time, simply repeat the 'Calibration' submenu several times.

Since it is necessary to know da / dN data in preparation for the SIF range in order to be suitable for some method, it is the same as the second task.

Computed da / dN against the fitted parameters and the SIF range compared to the input test data is stored as a rate-data-file name with a .dat extension.

4. Calculate fatigue life under various fatigue load spectra based on the crack growth (fracture mechanics) analytical approach. Various load interaction models can be implemented and used to suit different conditions.

To calculate the fatigue life, first follow all the necessary steps for the third task before entering the load data. Load specifications are entered from the load file by selecting the 'load-ready file' menu. Finally, the lifespan calculation is performed by selecting the 'crack growth' menu and the 'life calculation' submenu. Once all the necessary data has been prepared, the life calculation menu can be clicked.

5. Calculate fatigue life under various fatigue load spectrums based on crack initiation (traditional fatigue) analysis approaches, including SN curves and local strain life approaches.

In detail, first of all, select 'Start Crack' from the menu item, then select 'control parameters' and 'from file' from the submenu, and in response to the input request of the control-parameter-file name, In the menu item, select 'Start Crack' from the menu item, then select 'Material properties', 'from file' in the submenu, and select the property-data-file name. Enter the file name in response to the input request. Load specifications are entered from the load file by selecting the 'load-readyfile' menu.

Finally, life calculations are performed with the choice of 'crack initiation' menu and 'S-N life' or 'strain life' submenus. If all required data is not entered, the life calculation menus are inactive.

Hereinafter, a fatigue life prediction method of the present invention will be described in detail with reference to the accompanying drawings. 2 is a flowchart of a fatigue life prediction method of the present invention, FIG. 3 is a flowchart of a fatigue initiation prediction process, FIG. 4 is a flowchart of a crack growth analysis process, FIG. 5 is a flowchart of a life calculation process, and FIG. The main screen of the application program embodying the fatigue life prediction method of the present invention is shown.

As shown in FIG. 2, the present invention is divided into fatigue initiation prediction and crack growth analysis, and the fatigue initiation prediction is described first with reference to FIG. 3.

The fatigue initiation prediction largely includes a process of inputting control data, a process of inputting material properties, and a process of predicting S-N life or strain life according to a user's selection.

In detail, first, control data required for fatigue start prediction is input. The control data can be entered in advance and / or directly in the file. In other words, a file prepared in advance may be read or control data may be directly input by selecting control data from a menu, or a prepared file may be read and modified. The control data necessary for predicting fatigue onset is as follows.

1. Life curve: Choose from the modified life curve, S-N life curve, or modified S-N life curve.

2. When using the S-N method, an equal-life formula for evaluating the mean stress effect is selected from linear and parabolic curves.

3. Cycle counting method

4. Enter the number of different types of flights blocks.

5. Enter the plies (or cycles for the block loading case) within one spectral period.

6. Interpolation data indication: Enter 0 for no interpolation data, 1 for interpolation data for strain-N curves, or 2 for interpolation data for periodic stress-strain curves.

7. Number of Interpolation Data Points

8. The number of load-stress transformation states (coefficients).

Subsequently, material property values necessary for predicting fatigue onset using the strain life method or the S-N method are input. Of course, input data on material properties can also be directly inputted by file and / or.

When predicting fatigue initiation using the strain life method, the physical properties of the material are as follows.

1.fatigue strength coefficient (σ _f )

2. Fatigue ductility coefficient (ε _f )

3. fatigue strength exponent (b)

4. fatigue ductility exponent: c)

5. Young's modulus (E) or modulus of elasticity

6. strain hardening coefficient (K ')

7. strain hardening exponent (n ')

8. stress concentration factor (K _t )

9. Tolerances for stress-strain solutions (iteration convergency criterion, etc.).

In addition, when the fatigue start prediction using the S-N method, the physical properties of the material required are as follows.

1. Static strength (ST) used for equivalent life curve

2.S-N curve stress ratio (R)

3.S-N curve exponent (B)

4.S-N curve constant (A)

5. Young's modulus (E) or modulus of elasticity

6.S-N curve turning point stress (SB)

7.S-N limit lgN (S-N limit lgN: SN2)

8. residual error for equation solution (RSE)

9. Random number seed (AAA)

10.Scale factor for stress in S-N curve (TS)

11. Exponent in equal-life equation (QM). Where TS and QM are usually 1.0 and 2.0 in the equivalent lifetime equation.

As such, when both the control variable input and the material property input are completed, the SN life or the deformation life is predicted using the input data according to the user's selection, and the extension of .dat is added to other calculation results, that is, da / dN , Shape function β and the like are output.

The crack growth analysis will be described with reference to FIG. 3.

Crack growth analysis involves inputting geometric parameters to calculate the shape function used for stress intensity factor, inputting material properties to calculate da / dN, and loading conditions for life accumulation. It consists of the process of inputting the data and calculating the lifespan using the inputted data, and all the processes are composed of separate modules, so it is irrelevant to perform any one except when the lifespan calculation is performed.

In addition, the life estimation process includes selecting a type of a method to be applied to the crack growth analysis, calibrating elements in the selected method as needed, geometry parameters for life estimation, material properties, and If the load conditions are not entered, the method includes inputting them and performing a life calculation using the input data.

First, in the process of inputting the geometric parameters for calculating the shape function used in the stress intensity factor, when the necessary geometric parameters are entered, the shape function is automatically calculated for the crack length used in the stress intensity factor and has a .dat extension. It is stored as the geometry data file name.

The geometric parameters required for the calculation of the shape function are as follows.

1. a constant factor that provides a convenient coefficient for the geometric or load scale transformation applied during the calculation of β;

It is also required to input a geometric index to calculate the β factor.

When the values of the geometric indexes used are summarized as in Table 1 above, another index may be added in some cases, and '0' may be used as a mark indicating the end.

2. finite width (W), ie half the width of the SEN or the width of the CCT sheet metal;

3. thickness;

4. Initial crack length (a ₀ );

5. final crack length (a _f );

Geometric index value meaning One When using constant arguments in geometric parameters 2 When to use finite width correction 3 Β to a / w check using linear interpolation 4 Elliptic surface crack geometry 5 Compact specimen geometry (CT specimen) geometry (reference load is 'actual load') 6 Compact specimen geometry (CT specimen) geometry (reference load is 'actual load') 7 SENB (3-Point Bending) Specimen Geometry (Standard Span with S = 4W, Reference Load: Application Point Force) 8 Equivalent Stiffened Panels with Cracks Under Central Stringers 9 Table of Distributed Stress Tables (stress to x) 10 Stiffened panel with cracks between II (II by Poe) 11 Panels with non-destructive single stringers (III by Bloom) 12 IV by Bloom with a single breaking stringer; 13 Equal Steeppanel, Non-Destructive Central Stringer (I by Swift) 14 Equal Steeppanel, Non-Destructive Central Stringer (I by Swift)

6. semi-long axial of an elliptic surface crack (c);

7. distance between stringers (b);

8. Number of columns in the data table (NCOL) or number of stringers in half of the panel (NSTRG);

9. number of rows (points) in data table (NPTS) or number of stringers broken in half of the panel (NBROK);

10. stringer stiffness (EA) 2;

11. Young's Modulus (E1) on the panel; And

12. Stringer rivet distance (stringer fixed (rivet) distance: h).

Subsequently, when the material property data is input in the process of inputting the material property of the material for calculating da / dN, the da / dN value is automatically calculated in preparation for the SIF range. The calculated value is stored as a property data file name with a .dat extension.

The material property data required for the calculation of da / dN are as follows.

1. a (near) critical value of material toughness (Kc);

2. Threshold for ΔK (ΔKth);

3. Threshold (Kmaxth) for Kmax;

4. stress overload cut-off point (when using the Willenborg model and the value is 2-4): Rc);

5. Wheeler model index (m);

6. yield stress (σy);

7. rate formula coefficient (Paris, Forman & Walker): C);

8. rate formula exponent (all ways): n);

9. exponent associated with Kmax in Walker Model (q); And

10. C, n, q, etc. for the Walker equation with a negative R rate.

In addition, input of the number of stress ratios (NR) to the data of the crack growth rate is required. If there is no input to the crack growth data, NR can be set to '0'.

Subsequently, load conditions are input to calculate fatigue life. The load conditions required to calculate the fatigue life are as follows.

1. Number of start cycles (NZERO, usually 0);

2. The total number of flights in the flight-by-flight load spectrum (NFLITE) or the number of repetitions for the programmed block load spectrum (constant amplitude within the block).

3. Load Spectrum Type (LOADS)

(1) LOADS = 'MAXMIN' is a nominal load type and the load spectrum is specified by some load blocks under the conditions of maximum load and minimum load. The load block contains five numbers (SIGMAX, SIGMIN, CYCLES, NINT, NPRINT), which represent the maximum load, minimum load, number of cycles in the block, cycle integration interval and cycle result output interval.

(2) When LOADS = 'RANGE', the load spectrum specifies the conditions of the stress range and the periodicity R.

Each contains five numbers (TSUBA, R, CYCLES, NINT, NPRINT) for the load block, each of which has a double amplitude, block (sigma-min / Sigma max), and number of cycles in the block. , Cycle integration interval and cycle result output interval.

(3) When LOADS = 'FLBYFL', the load spectrum is again jointly specified as a condition by another flight consisting of a series of peak-valley points. Since the proximity to the highest-lowest point does not necessarily form a complete cycle, the user may make the following other choices.

First, the integer NG is read to indicate how to form a fatigue cycle.

NG = 1: Crack growth is estimated through all load reversals (a half cycle) assuming that reversal under load and under no load cause crack growth in the same way.

NG = 2: The crack is assumed to grow only during reversal under load, and the reversal under no load is completely ignored.

NG = 3: Fatigue cycle is counted by the Rain Flow Method.

Then, the number of different flight types (NTYPE) in the spectrum and the total number of flights (NFLGH) in all sections of the spectrum are read.

Finally, for each flight type, a file name (PN) containing the load data for the flight type and the number of iterations (IY) for that flight type across the spectrum are added.

After all the data required for fatigue life analysis have been entered, Life Accumulation is made according to the selected method, and the result is output.

The life calculation process can be performed even when all of the geometric parameters for calculating the shape function used in the above stress intensity factor, the material properties of the material for calculating the da / dN, and the load conditions necessary for the life calculation are entered. The geometrical parameters, the material properties and loading conditions for the calculation of da / dN can also be entered in the life calculation process.

Hereinafter, a process of calculating fatigue life will be described with reference to FIG. 5.

First, the method to be applied to the crack growth analysis is selected. Available options include Tabular, Paris, Forman, Walker, and Forman-Walker.

If there is an element that needs to be corrected, the correction is performed.

Elements that can be corrected include stress state, threshold, closure models, interaction models, etc. In stress state fixed plane, fixed plane ..., selfadjust st. .., reversedp ... can be calibrated, ΔK _th , Kmax thresholds can be calibrated at the threshold, Elber, Newman, Davidson, Chen-Lee and Elber-Chen-Lee models can be calibrated in the closed models, In the interaction model, you can calibrate the Wheeler models, Willenborg / Chang, HeQZ model, Chen-Lee, and Elber-Chen-Lee models.

The geometric parameters and material properties required for fatigue life analysis are the same as the geometric parameters in the shape function calculations and the material properties in the da / dN calculations described above, and if the geometric parameters, material properties and loading conditions are not entered in advance, This can be entered in the calculation, and once all the necessary work is in place, the corresponding lifetime is automatically calculated.

If the required data are not entered, no lifespan calculations are made. When the life calculation is executed, all the calculated results are saved in the form of output file. As mentioned above, in each analysis process, all input files (geometry, load condition, property value input, etc.) are saved with the extension .dat later. The first output file you specify stores all the results from each step. That is, the geometric input file stores the calculation results related to the geometry together with the generation of a file named geometric file name.dat, and the calculation result related to the load conditions is generated and stored in the load input file as the load file name.dat. All of these calculations are also stored in the first output file specified above.

Referring to the test results using the fatigue life prediction method of the present invention below, Figures 7a and 7b is a comparison result of the fatigue life using the fatigue life prediction method and the fatigue life prediction method of the present invention at a constant amplitude load, Figure 7a It can be seen that both the da / dN-ΔK curve and the aN curve of FIG. 7B are similar to the experimental results and the prediction using the fatigue life prediction method of the present invention.

8A and 8B are comparison results of fatigue life using the fatigue test under excessive / underload and the fatigue life prediction method of the present invention, and FIG. 8A shows the fatigue test and fatigue life of the present invention in which multiple overloads are applied. In the prediction method, the result of selecting the Chen-Lee model was compared and predicted, and FIG. 8B is a result of comparing the fatigue test with multiple over / underloads and the models of Wheeler, Willenborg, HeQz, Chen-Lee, and the like. The fatigue test results show that the existing Wheeler, Willenborg, etc. models are applicable to one overload, but Chen-Lee model has higher reliability under several more irregular overload / underload.

The fatigue life prediction method of the present invention as described above integrates the prediction of crack initiation and crack growth analysis into one, and integrates most of the existing crack growth models in crack growth analysis, and the crack growth models and their parameters are more consistent. Integrating da / dN test data processing (curve fitting) into an integrated package for simultaneous use allows simultaneous analysis of crack initiation and crack growth analysis, making it very convenient for predicting fatigue life of structures.

While the invention has been shown and described with respect to particular embodiments, it will be apparent to those skilled in the art that various modifications and variations can be made without departing from the spirit and scope of the invention as set forth in the claims. Anyone can see it easily.

Claims (5)

A computer-readable recording medium recording a program for predicting fatigue life of a structure, wherein the program predicts crack initiation and predicts the time when a crack occurs in a structure without defects, and crack growth analysis in which a structure's crack grows. Integrating, a) the process of inputting the control data and material properties required for the crack detection prediction by the crack initiation prediction, the selection of a method to be applied to the crack initiation prediction, and the computer processor by using the input data and the selected crack initiation prediction method. Calculating the SN life or strain life, and b) the method of applying the crack growth analysis to the geometric parameters for calculating the shape function used for the stress intensity factor, the geometric parameters for calculating the da / dN, and the load conditions for the life estimation, and the crack growth analysis. A computer program including a process of selecting a type of a crack and a crack initiation prediction result obtained in step (a), and performing a crack growth analysis with a computer processor using the input data and the selected crack growth analysis method. Recordable media that can be read by delete delete delete The method of claim 1, In the crack growth analysis, further comprising the step of correcting the input data and the type of the method to be applied to the crack growth analysis.