Reliable Life Prediction Research Articles

AI-assisted reliability life prediction model for wafer-level packaging using the random forest method

ABSTRACT We applied an artificial intelligence (AI) algorithm, the random forest (RF) model, using finite element simulations to predict the reliability life of wafer-level packaging (WLP). Due to rapid growth and an increasingly fast cycle pace of integrated circuits, it is imperative to shorten the development time of electronic packaging. This study focuses on packaging reliability analysis and prediction. In recent years, package reliability analysis has been performed using finite element method simulations, which reduce the required number of accelerated thermal cyclic tests. Compared with conventional ball grid array-type packaging, WLP has become the mainstream packaging type due to its small form factor, batch-type manufacturing process and low cost. We applied the RF model, a machine learning algorithm, to predict the reliability life of WLP. The finite element procedure, theory and mesh size were validated by a set of experiments, and a large dataset was generated for AI training purposes through the finite element simulations. The RF method was built using Python®. A fast and robust reliability assessment AI model of WLP can be achieved once the AI training accuracy is located within the target range; the designer only needs to input the geometry of each WLP component to obtain the reliability life cycle. WLP structural optimization can thus be easily achieved. The AI model also significantly shortens the design cycle to meet current design demands.

Study on creep-fatigue of heat sink in W/CuCrZr divertor target based on a new approach to creep life

Abstract Reliability and security of the fusion device are severely affected by the failure of the divertor. It is necessary to have a reliable life prediction for heat sink in W/CuCrZr divertor target, especially during long-term high heat flux operation in future fusion devices in which the high temperature creep accelerates its fatigue failure. In this study, a new approach to creep life was applied to creep-fatigue damage accumulation model for CuCrZr heat sink. It is proved that the results of new method for CuCrZr creep life calculation is more consistent with the experimental results than the previous empirical formula. Then the temperature and stress/strain results were calculated for two types of ITER-like monoblock using thermal–mechanical analysis which shows that high temperature creep could not occur on EAST divertor heat sink of current operating conditions. Based on creep-fatigue damage accumulation model, the predicted life of monoblock for DEMO divertor are given under different operating mode respectively. The model provides theoretical basis and guidance for the divertor engineering design of future fusion reactor like CFETR and DEMO in which the high temperature creep in heat sink must be taken into account due to long-term operation.

New theoretical framework for temperature-effect integration into asphalt concrete pavement life prediction with respect to Australian pavement conditions

The aim of this study is to introduce a new theoretical framework for incorporating the effects of temperature fluctuations into current pavement design approaches by using the cumulative damage concept. Until now, the effective temperature (Teff ) concept, which can integrate the effects of seasonal or monthly temperature fluctuations within a given annual temperature into pavement design, has been utilised, in conjunction with the dynamic modulus (|E*) master curve to determine the representative modulus for an asphalt layer in pavement design. The effective temperature concept is then used to aid the prediction of an asphalt concrete pavement’s service life by assuming that the effective temperature is constant throughout a pavement design period (approximately 20–40 years). However, this assumption is not particularly valid in some countries such as Australia, where the effect of climate change can cause significant fluctuations in temperature during a given pavement design period. In this newly proposed framework, the current effective temperature concept is modified to obtain a so-called optimum effective temperature (TOPT ). From this, the effect of temperature fluctuations during a given design period can be taken into account by using statistically based methods. First, a pavement design scenario was established to demonstrate the new concept. Historical climate data from four study areas in Western Australia was taken and a 10 mm dense graded asphalt mix was used as the basis for pavement design inputs with local material properties and traffic data. Results indicated that optimum pavement design sections, determined according to the new framework considering temperature fluctuations, provided thinner and thicker pavement sections with the cumulative damages (D value) closer to 1 than those of the traditional design approach. The study outcome would result in a more reliable pavement life prediction, leading to saving in construction and maintenance costs.

An Investigate on Degradation Models of Resonant Frequency and Mechanical Sensitivity for Butterfly Resonator Gyroscope

Analysis of degradation model of butterfly resonator gyroscope (BRG) is essential to predict its reliability. In order to establish the degradation model of the BRG, three parts of investigation are conducted. Firstly, an analytical model is developed to describe the effect of residual stress on the resonant frequency of excitation mode for BRG. Modal analysis of the BRG under different residual stresses (29 groups) is also simulated by finite element method (FEM), and the FEM results are close to the proposed analytical model within the range of residual stress (−24 MPa to 70 MPa). Secondly, the degradation models of resonant frequency at excitation mode and mechanical sensitivity of BRG are developed based the analytical model. Finally, the long-time static test of four BRGs at same environment temperature is designed and carried out. The test results show that the resonant frequency and mechanical sensitivity of the BRG decreases gradually with time increasing, which is consistent with the results of numerical simulation. Based on the degradation model of mechanical sensitivity, the mechanical sensitivity would degrade by 20.27% in 10 years. This work contributes to reliability evaluation and life prediction of BRG. [2019-0202]

The utility of laser scanning welds for improving fatigue assessment

Welded joints are a common source of fatigue failures. Traditional fatigue assessment approaches usually describe the weld with an idealized geometry defined by the weld flank angle and the weld toe radius. By doing so, the actual stress concentrations are not considered in such fatigue life predictions. 3D laser scanning technology offers an opportunity to more accurately represent the weld geometry, and accordingly perform fatigue assessment on more accurate representations of weld geometry. This manuscript explores the benefit of incorporating scanned weld geometries into linear elastic stress analysis for fatigue life prediction. Additionally, the capabilities to explain experimental scatter of six different fatigue indicator parameters have been quantified. Using experimental data from 28 tests across three geometries, a total of 60 deterministic and probabilistic fatigue models were assessed using both scanned and idealized weld geometries. The use of scanned weld geometries is found to substantially improve fatigue prediction when using local fatigue indicator parameters. However, if nonlocal fatigue indicator parameters are employed, the use of scanned weld geometries does not display a benefit. In total, the use of nonlocal fatigue indicator parameters with idealized weld geometry offer the greatest accuracy across the three cases examined. Thus, for linear elastic fatigue life assessment, the use of a nonlocal parameter in the fatigue life model is more motivated than utilizing 3D scanned weld geometry. Regardless of how the weld geometry is included, the signed von Mises equivalent stress led to the most reliable fatigue life predictions overall for the specimens considered in this work.

The Failure Evaluation and Reliability Life Prediction Technique of Complex Artillery Equipment

To accurately predict and master the actual combat capacity of the weapon system, the paper took a self-propelled artillery as the research subject and built the completely artillery system virtual prototype based on the dynamic theory. Extracted the dynamic characteristics of firepower system, running system and drive system, the performance variation and failure mechanism of the artillery system were studied by using the method of theoretical analysis, simulation and physical experiment. The paper built the fault evaluation and life prediction model in the process of shooting and drivingby combining data fusion, deep learning, etc. The performance evaluation and the prediction of fault of the full artillery system were realized. It can provide references for the life evaluation and prediction of complex equipment system.

Fatigue life prediction of spot-welded joints with a notch stress approach

Spot-welded joints are widely used in the construction of vehicle structures and frequently become critical locations for fatigue failure. Hence, it is essential to have reliable fatigue life prediction method for the spot-welded joints during vehicle structure design. In this paper, a new notch stress approach is developed for fatigue life prediction of the spot-welded joints. Currently, structural stress methods are widely used in automotive industry for fatigue life prediction of spot-welded joints. However, these methods are not well considering local geometry information. This paper introduces a notch stress based method to overcome the limitation of the structural stress methods. In the notch stress method, stress concentration factors for spot-welded joints are calculated from stress intensity factor equations. Then, the notch stress method is validated with fatigue test results of lap-shear and coach peel specimens.

Data-based models for fatigue reliability assessment and life prediction of orthotropic steel deck details considering pavement temperature and traffic loads

Considering the effects of long-term moving vehicle loads and ambient temperature on the fatigue life of details in orthotropic steel decks in cable-stayed bridges, a novel fatigue damage regression analysis model was developed and then used with reliability profiles to predict fatigue life. The regression model can characterize the normal correlation pattern between the daily averaged pavement temperatures, the annual average hourly aggregated traffic volume (AAHTV) or annual average hourly traffic weight (AAHTW) and a stress-based performance indicator. Prediction models based on the monitoring outcomes from the structural health monitoring systems installed on a Yangtze River cable-stayed bridge were first developed and used to describe and predict the time-series of the long-term moving vehicle loads and asphalt concrete (AC) pavement temperatures. An exponential regression model was then developed to quantify the relationship between daily averaged pavement temperatures, moving vehicle loads and the stress-based indicator, which is derived using S–N principles from strain measurements. Further, a limit equation based on the exponential regression model mentioned above for the fatigue life prediction, which considers both the traffic conditions and pavement temperature, has been developed. Finally, this methodology was employed to predict the fatigue life of details in orthotropic steel decks.

Life prediction of electric spindles with multiple degradation processes

In order to realize product life prediction without burst failure information, a method combining performance degradation information correlation analysis to construct a degradation process model is proposed. On the basis of obtaining the degradation information and establishing the single performance indicator degradation process model, the covariance is used to analyze the correlation of the multi-performance index degradation process, and the Copula connection function is introduced to establish the system reliability model under multidegradation process. Achieve the reliability curve drawing and reliable life prediction. Finally, the effectiveness of the method is verified by taking the degradation information of a series of milling type electric spindles as an example.

Improved backward fatigue statistical inference to the high-speed railway vehicle components

Original backward statistical inference method (BSIM) was improved (here further termed as ISIA) to fit the fatigue probabilistic S-N (P-S-N) curve with finite life data or small sample problem. Based on the search path, the novel scheme captures the optimal parameters that can efficiently establish the quantitative relationship between the life distribution and stress levels. Meanwhile, the equivalent lives at different stress levels can be obtained through the conversion of fatigue life. To formulate the more realistic P-S-N curve of full-size components, the life distribution effect of small-size specimens can be considered when fitting curves. Results show that the slope and intercept of newly-proposed fatigue P-S-N curve by using the ISIA for (X-x-x-x) type life data exhibit a relative error less than 3%, and the predicted fatigue life is approximately 5% of the group method (TGM). On the other hand, the predicted life is only 0.1% of (x-x-x-x) type life data by the classical method (TCM). For the welded structures widely used in high-speed railway vehicles, the present ISIA can achieve more conservative and reliable life predictions. It is thus concluded that the newly-proposed ISIA can not only ensure the accuracy of life distribution particularly during high cycle fatigue regime but also acquire more conservative P-S-N curve as opposed to standard methods such as TGM and TCM.

A notch stress method for fatigue life prediction of spot-welded joints

Spot-welded joints are widely used in the construction of vehicle structures and frequently become critical locations for fatigue failure. Hence, it is essential to have reliable fatigue life prediction method for the spot-welded joints during vehicle structure design. In this paper, a new notch stress approach is developed for fatigue life prediction of the spot-welded joints. Currently, structural stress methods are widely used in automotive industry for fatigue life prediction of spot-welded joints. However, these methods are not well considering local geometry information. This paper introduces a notch stress based method to overcome the limitation of the structural stress methods. In the notch stress method, stress concentration factors for spot-welded joints are calculated from stress intensity factor equations. Then, the notch stress method is validated with fatigue test results of lap-shear and coach peel specimens.

Research on Nonlinear Dynamic Simulation and Fatigue Reliability Life Prediction for Synthesis Transmission System of Self-Propelled Gun

Aiming to the puzzle that the inner load of nonlinear synthesis transmission system is difficult to obtain, a new kind of virtual prototype establishment and simulation method is put forward. The influence on nonlinear vibration with flexible rotor, bearing backlash is analyzed based on a virtual prototype. To validate the virtual prototype of nonlinear gear transmission system, the corresponding test platform is established. The consistency between simulation results and test results proves that the simulation results of the virtual prototype can be used to calculate the fatigue reliability life of key components. A new kind of fatigue reliability life prediction method of gear system considering multi-random parameter distribution is put forward based on the fatiguestatistic theory. Considering the periodicity of gear meshing, linear interpolation method is adopted to obtain the stress-time course of random load spectrum based on the gear’s complicated torque provided by virtual prototype. The gear’s P-Sa-Sm-N curved cluster can be simulated based on material’s P-S-N curve. The simulation process considers the parameter distributions of stress concentration coefficients, dimension coefficients and surface quality treatment coefficients, and settles the puzzle that traditional test methods cannot acquire the gear’s fatigue life of all reliability levels. This method can provide the distribution function and the interval of fatigue reliability life of gear’s danger region, and has a guide meaning for the gear maintenance periods determination and reliability evaluation.

Compilation of load spectrum for NC machine tool’s feed system based on the mixture weibull distribution

The load spectrum of the feed system for the numerical control (NC) machine tool is the objective basis of the reliability design and life prediction. So, a compilation method of load spectrum is proposed by using the mixture Weibull distribution (MWD). First, the load distribution in each feed direction is fitted by using the MWD. Among them, the Bayesian information criterion (BIC) determines the number of mixture components, the parameters of MWD are estimated by using the improved expectation maximization (IEM) algorithm, where the initial values of parameters are determined by the hidden Markov model (HMM). Second, the correlation between the loads in different feed directions is analyzed, and the joint distribution function (JDF) is established. Finally, the program loading spectrum is compiled by using the ratio coefficient. In this paper, the relative feed speed spectrum in the X-direction and Z-direction of NC lathes are established, case analysis results show that the relative feed speed distributions that use the proposed method have high fitting precision, and the load spectrum are compiled, which will be able to apply the reliability test, accelerated test and life prediction. Thus, the study of load spectrum has important theoretical significance and engineering application value.

Low-cycle fatigue characterization and texture induced ratcheting behaviour of forged AZ80 Mg alloys

Stress-controlled uniaxial “push-pull” fatigue testing was conducted on as-received (cast and extruded) and closed-die cast-forged and extruded-forged AZ80 Mg alloy. The as-cast material possessed random texture and somewhat symmetric cyclic responses. The extruded and forged materials possessed sharp basal texture and asymmetric cyclic responses. All materials exhibited tension/compression asymmetry in their cyclic response to varying degrees, depending on the thermomechanical processing conditions. It was discovered that the style of closed-die forging being investigated had spatially varying properties with texture orientations which varied based on the local forging directions and intensities which were dependent on the starting texture as well as the thermomechanical history. Under fatigue testing, the materials all developed some form of mean strain, with the nature and magnitude of this mean strain being dependent on primarily its texture intensity and propensity to twin in either tension or compression reversals. The type of mean strain (tensile or compressive) depends upon both the orientation and intensity of the starting texture of material. The texture induced ratcheting and resulting mean strain evolution was most pronounced in the as-cast material and had a significant impact on the fatigue life. Following forging, the material exhibited an increase in fatigue life of anywhere from 2 to 15 times for the cast then forged material and more modest yet still significant 8 times longer at stress amplitudes around 140 MPa for the extruded then forged material. The extruded forged material exhibited similar fatigue lives to that of the base material at stress amplitudes which approached the yield strength. The nature of the mean stress development and degree of fatigue life improvement depended on the processing conditions and the type of base material (cast or extruded) utilized to create the forging. Two energy based models were utilized to predict the life of the forged material, and gave a reliable life prediction for a variety of material conditions that were investigated.

A NEW PREDICTION APPROACH FOR THE STRUCTURAL FATIGUE LIFE BASED ON MULTI-FACTOR CORRECTION

As the phenomenon of fatigue damage is a common failure mode of equipment, the reliability evaluation and life prediction have become a hot-spot. The precise prediction of fatigue life in the initiation stage has become necessary. The common prediction study for structural fatigue life takes less influence factors into consideration. The common prediction results of fatigue life cannot be quantitatively corrected by the influence factors at the same time. This paper presents a research on the prediction approach for structural fatigue life based on the multi-factor correction. The influence of some factors on the fatigue life was analyzed and the prediction approach for structural fatigue life based on multi-factor correction was raised. Then the fatigue life of axle housing was predicted by using the corrected approach, as the case study. Moreover, the result predicted by the local stress–strain approach and the experimental data were used to verify the accuracy of the corrected approach. It can be clearly demonstrated by the results that the corrected prediction approach can be used to achieve the precision fatigue life for engineering structures. Further, it is also a prediction approach endowed with engineering application prospects.

Compilation of NC lathe dynamic cutting force spectrum based on two-dimensional mixture models

The load spectrum of numerical control machine tool (NCMT) is the basis data of reliability design, life prediction, and reliability bench test with simulated actual load conditions. Static cutting force spectrum, which can be compiled by empirical formula, cannot accurately reflect the real load conditions. Hence, a method for compiling the spectrum of dynamic turning force is proposed by using two-dimensional mixture models. Dynamic turning force is measured in the laboratory based on a large amount of cutting process data collected in the user field. Mean–frequency and amplitude–frequency matrices of dynamic turning force are obtained by the rainflow counting method. The probability distributions of mean and amplitude are established by using Weibull mixture models (WMM) according to the characteristics of the dynamic turning force, and the two main problems of this method are the determination of the number of mixture components and the estimation of parameters. Gray relation analysis is used to quantify the multi-objective evaluation matrix, and thus the number of mixture components is determined. Parameters are estimated using the improved expectation maximization (EM) algorithm, where the initial values of parameters are obtained by the K-means++ and Bayesian random classification. Finally, the joint probability distribution function between the mean and amplitude is established by the Copula function. Thus, the dynamic cutting force spectrum is compiled. Case analysis results show that the mean and amplitude distribution models with high fitting precision can be obtained using the proposed method. Furthermore, the precision of the load spectrum of NCMT is improved.

Interval statistic based reliability analysis method on small sample hot test of satellite thruster

• Propose an interval statistic based reliability analysis method fulfilling the life information from the failure time to the censored time. • Improve the reliability assessment accuracy of satellite thrusters obviously. • Conducted the coating test under different thermal conditions. • Conducted reliability assessment and life prediction of the satellite thruster in both transfer and synchronous orbit phases. As one of the most important components of satellites, the thruster must maintain high reliability, but the assessment is difficult when lacking test failure data. When conducting hot test of the satellite thruster, we encountered the problem of reliability assessment on type-I censored data with only one failure, which is common for small sample tests. This paper proposes a novel interval statistic based reliability analysis method, which can fulfill the life information from the failure time to the censored time ignored by conventional methods, and improve the assessment accuracy. In this paper, a life distribution model is established by proper failure mechanism analysis and prior test information exploitation. As thrust chamber burning-through has been considered as the main failure mode, life test of the chamber coating was conducted under different thermal conditions to obtain the model parameter. Based on the interval statistic theory, a detailed derivation is illustrated. Then, reliability assessment and life prediction for the satellite thruster in both transfer orbit phase and synchronous orbit phase have been achieved, and the results show that our method performs very well, which provides an important way for dealing with test data with only one failure.

Monotonic and cyclic behaviour of cast and cast-forged AZ80 Mg

Tensile and strain-controlled fatigue tests were performed to investigate the influence of forging on the performance of cast AZ80 magnesium alloy. The obtained microstructural analysis showed that the as-cast AZ80 magnesium alloy has dendritic α-Mg phase with eutectic Mg17Al12 morphology and a random texture. In contrast, the forged samples showed refined grains and a strong basal texture. During tensile testing, a maximum yield and ultimate tensile strength of 182MPa and 312MPa were obtained for the forged samples, representing increases of 121% and 33%, respectively, from the as-cast condition. At the same time, a significant improvement (73%) in ductility was obtained in forged samples. It was also observed that the forged samples achieved comparatively longer fatigue life under strain-controlled cyclic loading. Analysis of the fracture surfaces showed that a cleavage-type morphology was typical for the as-cast samples, while the occurrence of dimples and other evidence of plastic deformation were identified in the fracture surfaces of the forged specimens, indicating a more ductile response. Forging caused grain refinement and texture modification, both of which enhance alloy performance by improving strength and ductility, and leading to longer fatigue life. Strain and energy-based models were investigated for their suitability to predict the life of the forged material. Both the Smith-Watson Topper and the Jahed-Varvani energy-based models gave reliable life prediction.

The role of small specimen creep testing within a life assessment framework for high temperature power plant

ABSTRACTThe safe operation of ageing materials and structures operating at high temperature and pressure is challenging. This paper, for the first time, clearly addresses the role of small specimen creep testing methods within a practical life assessment framework, with a case study used to illustrate the main principles and applications. To enhance the current practice for assessing the condition of creep ageing components, more proactive use of small specimen testing methods is proposed for the in-service condition assessment of power plant materials, earlier in the plant lifecycle and within a holistic life assessment framework. The paper describes how small specimen creep testing methods and other complementary tools can be used together in a new structured approach to life management. There is a clear need to provide the plant owner with more reliable and effective life prediction tools, based on earlier and more rigorous assessment of in-service life consumption. A novel, holistic, life assessment methodology including the use of small specimen creep testing has been developed by the authors.

A fatigue damage and residual strength model for unidirectional carbon/epoxy composites under on-axis tension-tension loadings

Fibre-reinforced composites experience a degradation of stiffness and strength during fatigue life. Understanding the reduction of these properties is fundamental to establish a reliable fatigue life prediction methodology. This work investigates the loss of stiffness and strength in advanced unidirectional carbon/epoxy laminates under on-axis tension-tension loads. A phenomenological stiffness-based fatigue model is formulated within the framework of continuum damage mechanics, where damage is described by the reduction of the in-plane longitudinal stiffness. The particularity of the model is to assume that the ultimate strain remains constant after fatigue damage. Thus, the residual strength model and the S-N curves are deduced from the residual stiffness model. This assumption reduces the experimental characterization of phenomenological-based approaches. The experimental challenges found in the fatigue experiments are also discussed. The accuracy of the model is verified by comparing the experimental data with the derived S-N curves.