Methods for simulation-based emission of operational load signals

Abstract

The invention relates to a method for simulation-based emission of operating load signals to shorten fatigue strength tests, in particular for multi-channel loaded components, characterized by 1. Local stress calculation using the finite element method (FEM) and 2. Local fatigue damage calculation and identification of relevant reversal points, consisting of the following sub-steps: 2.1. Creation of a reference voltage-time curve, 2.2. Indexed Rainflow Counting and Classification, 2.3. Damage accumulation. 3. Reconstruction of the load-time profiles based on the relevant time points

Description

[0001] The efficient and reliable omission of operational load signals to shorten fatigue strength tests presents a significant challenge, particularly for components subjected to multi-channel loading. A novel omission method is presented, based on FEM-based fatigue strength simulation and capable of handling any number of load channels. A specialized omission algorithm first modifies the processing of the rainflow matrix and identifies the necessary turning points in the stress-time profiles. From these, new operational load signals are reconstructed. To account for the typically more complex shape of the stress spectrum resulting from omission, a correction for the test duration is proposed.

[0002] The presented method is illustrated using the example of a support arm subjected to four load channels.

[0003] To this day, the approval of automotive chassis components for series production is based on experimental fatigue tests. With regard to saving time and costs, shortening the testing time is desirable, and the concept of omission as a reduction method is preferred, since the exclusive omission of low-amplitude load cycles underlying this concept has the least impact on the test result. In contrast, methods in which the reduction of the number of cycles is achieved, for example, by increasing the load, can influence the test result. Fig. Figure 1 shows the basic application of omission to a load-time curve.

[0004] For a profitable omission procedure, there are fundamentally two opposing requirements regarding the intensity of omission (number of skipped swings): 1. Increase the intensity of omission as much as possible to achieve maximum time and cost savings (efficiency) 2. Intensity of omission low enough to keep the influence on the test result negligible (reliability)

[0005] Components subjected to multiple and typically poorly correlated load-time signals present a particular challenge with regard to these requirements, for which no generally applicable methodology exists to date. The in Fig. Figure 2 shows that the control arm of a passenger car's rear axle, when mounted on three points, is subjected to four different load-time profiles. A method is presented below that allows for efficient and reliable emission testing while adequately considering all load-time profiles.

[0006] According to the state of the art, the following measures are possible to reduce test times according to [1] (see also Fig. 3): a) Increase in load change frequency - limited by testing technology, often little potential for reducing testing time in practical application - not readily applicable to all materials and also not applicable to time-dependent processes (e.g. corrosion, creep) b) Removal of stress breaks - Removal of holding times in the load-time curve; in practical application, often little potential for reducing test time. - not readily applicable to all materials and also not applicable to time-dependent processes (e.g. corrosion, creep) c) Increased workload - proportional enlargement of all amplitudes - limited by testing technology, risk of changes to damage mechanisms and crack initiation positions (e.g., due to introduced plasticization), distortion of the influence of residual stresses - not readily applicable to all materials and also not applicable to time-dependent processes (e.g. corrosion, creep) d) Increase in the fullness of the load collective - Application of a larger number of swings with larger amplitudes - Risk of changes to damage mechanisms and crack initiation positions (e.g., due to introduced plasticization), distortion of the influence of residual stresses - not readily applicable to all materials and also not applicable to time-dependent processes (e.g. corrosion, creep) e) Truncation - Omission of individual, rare load peaks if these introduce compressive stresses at fatigue-critical points. - conservative shortening of the service life and, in a narrower sense, falsification of the test result f) Omission - Omit swing games with small amplitude, as these contribute very little to damage. - Very efficient method - Determining the omission level (limit of the amplitude below which omission occurs) is, however, a complex problem. - Omission based on the load collective is not always suitable due to sequence effects. - not readily applicable to all materials and also not applicable to time-dependent processes (e.g. corrosion, creep)

[0007] Due to its efficiency and the greatest potential for reducing test time, omission proves to be the preferred method and is the subject of the present invention, wherein the chosen methodological approach solves the particular difficulty of determining the omission level while preserving sequence effects. The widely used textbooks on fatigue strength reveal little insight into this topic.

[0008] In [2] a general emission level (defined here as the ratio of the maximum stress amplitude which is omitted to the fatigue strength) of 50% is recommended.

[0009] In [1] this order of magnitude is confirmed for steel components, provided that the load cycles are corrected for mean stress.

[0010] Today's widely used omission methods, which are also included as standard tools in software applications for measurement signal and data processing, generally focus on directly processing load-time histories. This only partially addresses local fatigue processes on the component, as the relevance of individual load channels to fatigue-critical areas is not evaluated. The common use of "virtual" S-N curves also appears disadvantageous for achieving the requirements of omission (efficiency and reliability), since the primary S-N curve parameter relevant for omission—the slope—is only assumed in general terms, but is actually also significantly influenced by the local stress situation on the component (e.g., notch sharpness).

[0011] A literature review on practical application examples of emission methods mainly yields results in the field of aerospace.

[0012] In [3], aluminium material samples are investigated with a focus on fatigue life (including crack growth), and the achieved reductions in test time as a function of the emission level are presented. The investigations are limited to single-channel loading and are not transferable to multi-channel loading.

[0013] The omission on composite structures is described in [4] and [5], where the underlying methods also appear to be possible only for single-channel loading.

[0014] The disadvantage of classical omission methods, which often cannot provide sufficient reduction in test time, is discussed in [6] and a “smart compression” is added as a new method.

[0015] DE 44 18 599 A1 discloses, for example, a method for simplifying measured, multiaxial load-time functions for fatigue tests of components on testing machines with servohydraulic cylinders, comprising the steps of: calculating resulting load-time functions of uniaxial component stresses from measured, multiaxial load-time functions; determining those reversal points in the time functions of the uniaxial component stresses that belong to non-damaging load cycles and marking all other (damage-relevant) reversal points in the time functions of the component stresses; marking all reversal points and transients of the load-time functions that occur simultaneously with the (damage-relevant) reversal points marked in all time functions of the component stresses; reducing the load-time functions to exclusively these reversal points and transients, which form the support points;Calculation of the condensed load-time functions by calculating transition functions between the support points and taking into account the maximum values ​​of the velocity, acceleration and frequency of the testing machine.

[0016] German patent application DE 10 2019 216 842 A1 discloses a method for determining the service life of an electric drive machine, the method comprising the steps of: acquiring load information that indicates a load on the drive machine; determining a load collective based on the load information; deriving a stress on the drive machine based on the load collective; providing stress resistance information that indicates the stress resistance of a material of at least one component of the drive machine, and comparing the load and the stress resistance information to determine the service life of the drive machine.

[0017] The use of numerical optimization methods is also described in

[11] .

[0018] In summary, it can be stated that currently no emission testing methods are available that equally meet the requirements for efficiency and reliability, especially for multi-channel loaded components, so the object of the invention is to provide such a method that meets these requirements.

[0019] To solve this problem, the invention proposes the following: Meeting the requirements for efficiency and reliability necessitates considering the local fatigue processes within the component. Accordingly, the numerical methods currently used for fatigue strength analysis / service life calculation provide a suitable basis. These typically consist of two analysis steps (see also Fig. 4): 1. Local stress calculation using the finite element method (FEM) 2. Local fatigue damage calculation The second step of the analysis, the local fatigue damage calculation, is subdivided into three essential sub-steps (see also Fig. 5), which must be processed for each FE node to be analyzed and (when using a section plane method) for each section plane: 1. Creation of a reference voltage-time curve 2. Rainflow counting and classification 3. Damage accumulation, whereby for each FE node (damage hotspot) of the component to be analyzed, the necessary reversal points of the local equivalent stress-time history present at the respective node are to be determined by indexed rainflow counting, which lead to the essential damage contribution, whereby a set of reversal points is available at each FE node after processing, wherein the union set resulting across all FE nodes to be analyzed contains all reversal points which are required for the creation of the new load-time histories, whereby for each occupied element of the rainflow matrix the partial fatigue damage values ​​are first calculated conventionally, which are then summed to determine the total damage, and during the omission process a second pass of all occupied elements of the rainflow matrix is ​​then carried out.However, in ascending order of the mean-stress-corrected stress amplitudes and for each matrix element, a decision is made based on the respective partial damage value as to how many of the respective cycles can be omitted, so that overall – across all matrix elements processed up to that point – no more than a previously defined proportion of the total damage is lost, whereby the remaining – not omitted – cycles are used to determine the reversal points required at this node with the help of the indexing of the Rainflow matrix.

[0020] To best account for local fatigue processes and the local influence of individual load channels, and to ensure efficiency and reliability, the omission is no longer performed based on individual load-time histories, but now on the basis of the comparative stress-time history. Accordingly, the execution of an analysis process according to Fig. 4. Prerequisite for the novel emission procedure.

[0021] The corresponding modeling and calculation effort can be considered acceptable, since appropriate simulation models already exist within the scope of component development, or these can be generated and calculated relatively quickly. Furthermore, with regard to the objective of omission (an accurate calculation of all fatigue damage values ​​in detail is not absolutely necessary for this purpose), model simplifications are possible, particularly in the area of ​​interfaces.

[0022] To identify the damage-relevant time points, the necessary reversal points of the local equivalent stress-time history at each FE node (damage hotspot) of the component to be analyzed must be determined. These reversal points must contribute significantly to the damage. After processing, a set of reversal points is available at each FE node. The resulting union across all FE nodes to be analyzed then contains all reversal points required for generating the new load-time histories. The basis for determining the reversal points at a single FE node is the so-called indexed rainflow count. This extended counting method has already been applied in [7] to obtain sequence effects. In this method, in addition to the number of cycles, the two relevant time points for each cycle are also stored in the rainflow matrix for each element (see also [7]). Fig. 6. For each occupied element of the Rainflow matrix, the partial fatigue damage values ​​are first calculated conventionally and then summed to determine the total damage. During the omission process, a second iteration of all occupied elements of the Rainflow matrix is ​​then performed, this time in ascending order of the mean stress-corrected stress amplitudes. For each matrix element, a decision is made based on the respective partial damage value as to how many of the cycles can be omitted, ensuring that no more than a predefined percentage of the total damage (e.g., 5%) is lost overall (i.e., across all matrix elements processed up to that point). Using the indexing of the Rainflow matrix, the remaining (not omitted) cycles can then be used to determine the reversal points required at that node.

[0023] The required reversal points based on the comparative stress-time history represent the support points for the subsequent reconstruction of the load-time histories.

[0024] Sometimes clusters of hotspots remain. These are neighboring hotspots that represent nearly identical load values ​​across load channels. These clusters can arise from considering more than a single hotspot during omission, if the damage-relevant hotspots have a slight time offset. Possible causes for this include...

[0025] Superposition effects for the individual channels, as well as dynamic effects under higher-frequency loads (inertial effects), are considered. Since clusters can introduce noticeable discontinuities into the reconstructed load-time profiles, these are identified and removed in a correction step ( Fig. 7) This may have a minor but practically irrelevant effect on fatigue damage values.

[0026] To reconstruct all load-time profiles (reconstruction of the operational load signals), interpolation is performed between each pair of data points using a cubic polynomial. The four parameters of the cubic polynomial are determined from the position (function value) of the two data points and the tangents (first derivative) at these two data points. While the function values ​​at both data points are derived from the original load-time profiles, the tangents are initially undefined. Using a specific iterative algorithm, the tangents are determined such that all reconstructed load-time profiles are continuous over the entire time domain up to the second derivative. Thus, the reconstructed operational load signals are not only tangent-continuous (without kinks) but also curvature-continuous. These continuity properties are advantageous for experimental testing and, in particular, for the test bench iteration required beforehand.

[0027] One disadvantage of the described interpolation method is that local extrema sometimes occur between the support points; see also Fig. 8.

[0028] Since the support points correspond to the damage-relevant inflection points in the stress-time histories and thus, in almost all cases, also represent the inflection points in the load-time histories, ideally only the support points should represent the local extrema in the reconstructed load-time histories, as otherwise an "artificial" increase in fatigue damage can occur. This can be remedied by correcting the corresponding polynomial segments (adjusting the slopes at the polynomial edges), thereby reducing the local extrema between the support points to a definable tolerance threshold. While this leads to a loss of curvature continuity at the edges of the corrected polynomial segments, tangent continuity is always maintained.

[0029] A complete reduction of local extrema (zero tolerance threshold) is not favored here, as the reconstructed load-time histories then exhibit very high local curvatures in some cases, which are visibly reflected as kinks (although the polynomials remain mathematically continuous along tangents). The same problem arises if the tangents (first derivative) at the support points are set to zero from the outset, which is why this approach was rejected early on. A low chosen tolerance threshold, and thus the acceptance of even minor remaining local extrema, proves to be a good compromise between minimizing the "artificial" increase in damage and achieving the smoothest possible (i.e., kink-free) reconstructed load-time histories.

[0030] The load-time signals reconstructed using the described interpolation are then sampled based on a predetermined frequency (see Fig. 9) and the resulting discrete load data in a suitable format (e.g. rpc) written to files which can be used for experimental testing. Consideration of the shape of the stress collective

[0031] Fatigue failure (e.g., in the form of a technical crack) is determined according to the original form of the miner's rule at the so-called theoretical damage sum D. th Assuming a value of 1.0, while the Wöhler test (constant amplitudes, single-stage loading) can still be accurately described, experience has long shown that fatigue failure in the Gassner test (variable amplitudes, multi-stage loading) often occurs even when a calculated lower total damage amount, the so-called actual total damage amount D, is reached. tat≤1.0 occurs. Among other things, sequence effects (temporal sequence of swings with large / small amplitudes) are considered to be the cause, which are not taken into account in the relatively simple damage accumulation calculations used to date.

[0032] Current calculation methods determine the actual amount of damage D tat depending on a parameter that describes the shape (or completeness) of the stress spectrum. In the FKM guideline [8], the horizontal distance A is specified as the parameter. ele The Wöhler curve is proposed as a conversion to the Gassner curve (service life curve), which can be calculated elementary from the stress collective (frequency distribution of the mean stress-corrected stress amplitudes) using the Miner variant. The actual damage sum D tat - in the FKM guideline synonymous with effective miner sum D m designated- can then depending on A eleare calculated, whereby these are limited downwards by a material group-dependent lower limit D m,min is limited, see Fig. 10.

[0033] The improved accuracy of computational lifetime predictions using D m =D tat The simultaneous application of the modification of the Miner rule according to Haibach (continuation of the Wöhler curve below the inflection point with the slope 2k-1) has already been shown in [9] using 293 steel samples.

[0034] Due to the omission and the associated loss of small swing cycles, the stress collective becomes more complete, the distance A ele between Wöhler line and Gassner line smaller and the actual damage sum D tat =D m larger.

[0035] To ensure that the fatigue strength test result is not distorted by the omission, the relationship between fatigue stress and fatigue strength must be maintained. The fatigue stress is represented by the calculated damage resulting from the load-time histories and, according to the assumptions, is not subject to any significant changes due to the omission. The fatigue strength is represented by the actual damage amount D. tat =D m represents, which is increased to some extent by the omission.

[0036] In conclusion, it is proposed that the duration of tests using load-time curves shortened by means of omission be reduced by the ratio of the actual damage amounts D. tat =D mto increase emissions both before and after. Note: To avoid confusion with the existing damage amounts resulting from operational loads (in the sense of fatigue stress), the term "actual damage amount D" is used instead. tat =D m The following is the designation "acceptable amount of damages D". m (in the sense of fatigue strength). Example:

[0037] A component is to be subjected to a fatigue strength test with a load-time curve that is to be repeated 10,000 times. The omission leads to an increase in the tolerable damage amount of D at the fatigue-critical point of the component due to the now more complete stress spectrum. m =0.5 on D m =0.6. Using the shortened load-time curve due to the omission, the test must be carried out with 10,000·1.2=12,000 repetitions according to the ratio 0.6 / 0.5=1.2.

[0038] Part of the reduction in testing time achieved through the Omission must be invested in this corrective proposal, although it can be expected that only a very small proportion of the reduction in testing time achieved through the Omission will be lost again.

[0039] An application example of the method according to the invention is explained below: Test scenario

[0040] The rear axle control arm shown in the introduction (see above) Fig. 2) is to be tested using a 4-channel operating load signal. The operating load signal has a duration of 34.4 s and is to be repeated 20,000 times, resulting in a total test duration of 191 hours (approx. 8 days) per component.

[0041] The emission procedure shown can be used to shorten the testing time.

[0042] This involves first performing a full-field damage calculation and identifying relevant damage hotspots, which form the basis for the omission. A maximum permissible loss of damage of 1% is applied as the basis for the omission. Result of the submission

[0043] The fatigue damage before and after the emission shows Fig. 11 and Fig. 12 and the table in Fig. 13.

[0044] The influence of omission on the damage levels at the hotspots is generally negligible.

[0045] At many hotspots, a slight increase in damage values ​​can be observed (initially counterintuitively), which is due to the fact that the interpolation of the load-time profiles occasionally results in local extrema between neighboring support points, which have not been completely reduced in favor of a more continuous load-time profile, cf. Fig. 8. Hotspot 1118283 is an exception, showing a significant 35% increase in damage. This is because all stress amplitudes at this hotspot lie below the inflection point of the S-N curve, meaning that a change in load results in the significantly flatter slope of the S-N curve as modified by Miner. However, this effect only occurs at comparatively very low damage levels and is therefore acceptable.

[0046] At other hotspots, there is a slight reduction in damage values, which in some cases exceeds the defined maximum permissible damage loss of 1%. The reason for this is again the distance between clusters of neighboring support points, cf. Fig. 7. To search. However, the losses due to damage remain within a completely acceptable range.

[0047] In Fig. Figure 13 also lists the von Mises equivalent stresses that occur at their maximum over time at the individual hotspots. As expected, the omission does not lead to any significant change in the maximum stresses. The slight stress increases are also due to the fact that the interpolation of the load-time profiles occasionally results in local extrema between adjacent support points, which were not completely eliminated in favor of a more continuous load-time profile (see Figure 13). Fig. 8. With regard to the yield strength of the material of 260 MPa, which is only slightly above the maximum von Mises equivalent stresses, this behavior is important because an increase in the maximum equivalent stresses - which would be the case with other methods for reducing test time - would induce residual stresses that could distort the fatigue strength behavior. Fig. 14 to Fig. Figure 17 illustrates the load-time profiles for the four channels before and after omission. It is important to note that the temporal interaction of all channels is fully preserved by the omission procedure.

[0048] Fig.Figure 18 shows the stress amplitude spectrum, exemplified by the most severely damaged hotspot. The omission level here is 72% (relative to the local fatigue strength). It is worth noting that above the omission level, there are very slight changes in the spectrum, which can be attributed to the removal of clusters of support points. However, this influence is not relevant and can be disregarded. It is also noteworthy that below the omission level, not all stress amplitudes are clipped as expected, and the typical vertical limitation in the spectrum is not observed. The reason for this lies in the mutual influence of the individual damaged hotspots – the "large" (damage-relevant) stress amplitude at hotspot A remains, in a sense, as a "small" stress amplitude at the adjacent hotspot B.If the omission were based solely on a single damage hotspot on the component, the "textbook" progression of the collective after the omission would also be present.

[0049] After applying the omission procedure, a 4-channel load-time signal with a duration of 3.5 seconds remains. With 20,000 repetitions, this would correspond to a total test runtime of 19.4 hours (approx. 0.8 days) per component and a 90% reduction in test time due to omission. However, as a result of omission, the acceptable damage amounts increase, to a maximum of approximately 1.8 times. Therefore, a corresponding increase in the number of repetitions to 36,000 is recommended, which increases the total test runtime to 35 hours (approx. 1.5 days). This still results in a significant test time saving of 82%.

[0050] Omission, as an efficient and minimally invasive method for reducing testing time, is procedurally integrated into today's established modern fatigue strength simulation methods. This is based on an extension of rainflow counting using indexing, which allows the identification of damage-relevant reversal points in the local stress-time histories at the respective damage hotspots. These damage-relevant reversal points represent the support points in the load-time histories newly generated through interpolation and sampling.

[0051] The main advantages of the presented emission procedure are: - Consideration of local mechanical stresses on the component - Possibility of considering any number of load channels, whose relevance is automatically weighted by the process. - Maintaining the interaction of loads from the individual load channels - Reconstruction for testing suitable (easily iterable) load-time signals that are tangent-continuous over the entire time domain and also curvature-continuous over large areas. - Specifying a maximum permissible loss of damage instead of specifying an emission level - No influence of omission on fatigue strength behavior, since ◯ the fatigue damage remains ◯ the maximum voltages are maintained ◯ the amplitude voltage collectives are largely retained (except for the omission of small amplitudes) - consideration of the increased collective completeness due to the omission and the resulting increased tolerable damage sums through an equivalent extension of the operating time ◯ the order effects are preserved

[0052] The necessary mapping of the component in a simulation model based on the finite element method (FEM) for the application of the omission procedure represents an acceptable additional effort. Experience shows that these simulation models are often already available as a result of the component development. Should the creation of a simulation model nevertheless be required, certain model simplifications (e.g., omitting the detailed modeling of interfaces) can be made for the purpose of omission.

[0053] The aforementioned advantages have been illustrated using the example of a 4-channel loaded rear axle control arm. The Omission promises a test time reduction of 82% without any significant impact on the mechanical behavior or fatigue behavior of the component.

[0054] The application of the omission method presented here in particular (or omission in general) is initially limited to metallic components whose fatigue strength behavior is largely independent of the loading frequency.

[0055] The application of omission to components exposed to corrosive loads in addition to mechanical ones (e.g., aluminum components exposed to salt water) remains subject to future investigation. It is conceivable that the corrosive medium could be made more aggressive in accordance with the shorter test time, as is also proposed, for example, in

[10] . literature [1] Vajen, Henning: “Investigation of the influence of practical testing conditions on the fatigue strength of components of the common rail diesel injection system”, Dissertation, Materials Testing Institute University of Stuttgart, 2014. [2] Radaj, D.; Vormwald, M.: „Ermüdungsfestigkeit - Grundlagen für Ingenieure“, 3. Auflage, Springer, 2007. [3] Hailing, Tian; et al.: „Influence of low load truncation level on crack growth for Al 2324-T39 and Al 7050-T7451“, Chinese journal of aeronautics 22.4 (2009): 401406. [4] Healey, Rowan; et al.: „The application of cycle merging and an extension of a fatigue spectrum simplification methodology from unidirectional to woven composite materials“, Composites Part C: Open Access 8 (2022): 100283. [5] Clark, G.; T. J. Van Blaricum: „Load spectrum modification effects on fatigue of impact-damaged carbon fibre composite coupons“, Composites 18.3 (1987), 243-251. [6] Wallbrink, Chris: „Smart load spectrum compression through the preservation of damage content“, 17th Australian International Aerospace Congress (AIAC17), Melbourne, Australia, 2017. [7] Healey, Rowan; et al: “A review on aircraft spectra simplification techniques for composite structures” Composites Part C: Open Access 5 (2021): 100131. [8] Rennert, R.; Kullig, E.; Vormwald, M.; Luke, M.: “FKM guideline, computational strength verification for machine components”, 7th revised edition, VDMA Verlag, 2020. [9] Hinkelmann, K.: “Correction functions for improving the computational lifetime estimation under cyclic loading”, Dissertation, Papierflieger Verlag, 2012.

[10] Sonsino, CM: “Reduction of test time in fatigue strength testing.” MP Materialprüfung, (45 / 4), 2003.

[11] SCHRANK, Ronald: Definition of simplified fatigue tests using numerical optimization. In: Procedia Structural Integrity, 2022, Volume 38, pp. 30-39.

Claims

[1] Method for simulation-based emission of operational load signals to shorten fatigue strength tests, especially for multi-channel loaded components, comprising the following analysis steps:

1. Local stress calculation using the finite element method (FEM) and 2. Local fatigue damage calculation and identification of relevant reversal points, consisting of the following sub-steps: 2.

1. Creation of a reference voltage-time curve, 2.

2. Indexed Rainflow Counting and Classification, 2.

3. Damage accumulation, 3. Reconstruction of the load-time histories based on the relevant time points, characterized in that, for each FE node (damage hotspot) of the component to be analyzed, the required reversal points of the local equivalent stress-time history present at the respective node are to be determined by indexed rainflow counting, which lead to the essential damage contribution, wherein a set of reversal points is available at each FE node after processing, wherein the union set resulting across all FE nodes to be analyzed contains all reversal points that are required for the creation of the new load-time histories, wherein for each occupied element of the rainflow matrix the partial fatigue damage values ​​are first calculated conventionally, which are then summed to determine the total damage, and subsequently, during the omission process, a second pass of all occupied elements of the rainflow matrix is ​​carried out.However, in ascending order of the mean-stress-corrected stress amplitudes and for each matrix element, a decision is made based on the respective partial damage value as to how many of the respective cycles can be omitted, so that overall – across all matrix elements processed up to that point – no more than a previously defined proportion of the total damage is lost, whereby the remaining – not omitted – cycles are used to determine the reversal points required at this node with the help of the indexing of the Rainflow matrix. [2] Method according to claim 2, characterized by , that clusters of support points arising during omission, which are to be understood as neighboring support points representing almost identical load values ​​across load channels, are identified and removed in a correction step. [3] Method according to any one of claims 1 to 3, characterized by, that to reconstruct all load-time profiles between any two support points, interpolation is performed using a cubic polynomial, and the four parameters of the cubic polynomial are determined via the position (function value) of the two support points and the tangents (1st derivative) at the two support points, whereby the function values ​​at both support points result from the original load-time profiles, and the tangents are determined using an iterative algorithm such that the entire reconstructed load-time profiles are continuous over the entire time domain up to the 2nd derivative. [4] Method according to claim 4, characterized by , that a correction of polynomial segments is made between support points between which local extrema occur, whereby the correction reduces the local extrema to a tolerance threshold to be determined. [5] Method according to claim 4 or 5, characterized bythat the load-time signals reconstructed by means of the described interpolation are finally sampled on the basis of a specified frequency and the resulting discrete load data are written in a suitable format to files which can be used for experimental testing.

Images