Reliability Simulations for Solder Joints Using Stochastic Finite Element and Artificial Neural Network Models

1996 ◽  
Vol 118 (3) ◽  
pp. 148-156 ◽  
Author(s):  
G. Subbarayan ◽  
Y. Li ◽  
R. L. Mahajan
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Finite Element ◽  
Solder Joint ◽  
Six Sigma ◽  
Manufacturing Process ◽  
Solder Joints ◽  
Finite Element Models ◽  
Process Tolerance ◽  
Reliability Of Solder Joints

The field reliability of solder joints depends on the manufacturing process tolerance of design parameters and on the capability of manufacturing processes to achieve the tolerance. This process capability is usually expressed through measures such as “six-sigma.” In this paper, a systematic procedure to estimate the reliability of solder joints due to manufacturing process induced variations on the design is presented. The reliability is calculated using the stochastic finite element method and is most naturally expressed in terms of a mean life and a standard deviation in life. An integrated finite element solution procedure for predicting solder joint profile (during reflow) and life is also presented in the paper. A physico-neural approach in which the finite element models are used to build an artificial neural network model is next developed to combine the accuracy of the finite element models with the computational efficiency of neural networks. This physico-neural approach is shown to reduce the computational time required per design evaluation by four orders of magnitude without significant loss of accuracy. The developed procedures are applied to the 72 I/O OMPAC BGA package from Motorola, Inc. It is shown that a ±10 percent process tolerance on solder joint height, volume and pad sizes with a “six-sigma” process capability on these variables will result in solder joint with over ±20 percent variations in life about the mean life at ±6σ level. It is also shown that variations in life of BGA solder joints are most sensitive to variations in solder joint height. Variations in PWB pad size, solder volume, and substrate pad size are relatively less important, but in the order listed.

Predicting Thermal Fatigue Lifetimes for SMT Solder Joints

1992 ◽  
Vol 114 (4) ◽  
pp. 472-476 ◽  
Author(s):  
J. Sauber ◽  
J. Seyyedi
Keyword(s):  
Finite Element ◽  
Solder Joint ◽  
Thermal Cycling ◽  
Thermal Fatigue ◽  
Solder Joints ◽  
Finite Element Models ◽  
Creep Tests ◽  
Creep Equation ◽  
Joint Behavior ◽  
Creep Strains

A power-law type creep equation has been added to finite element models to calculate solder joint response to time, temperature, and stress level. The ability of the models to predict solder joint behavior was verified by running a series of creep tests. The models were then solved to determine the solder joint creep strains which occur during thermal cycling. These creep strains were used to predict the degradation of pull strength resulting from thermal cycling. More than 8,600 solder joints were thermally cycled and then individually pull tested to verify the accuracy of the method.

scholarly journals Thermal field prediction in DED manufacturing process using Artificial Neural Network

2021 ◽  
Author(s):  
Seifallah Fetni ◽  
Quy Duc Thinh Pham ◽  
Van Xuan Tran ◽  
Laurent Duchêne ◽  
Hoang Son Tran ◽  
...  
Keyword(s):  
Neural Network ◽  
Machine Learning ◽  
Artificial Neural Network ◽  
Finite Element ◽  
Manufacturing Process ◽  
Thermal Field ◽  
Activation Function ◽  
Tungsten Carbides ◽  
Artificial Neural ◽  
The Impact

In the last decade, machine learning is increasingly attracting researchers in several scientific areas and, in particular, in the additive manufacturing field. Meanwhile, this technique remains as a black box technique for many researchers. Indeed, it allows obtaining novel insights to overcome the limitation of classical methods, such as the finite element method, and to take into account multi-physical complex phenomena occurring during the manufacturing process. This work presents a comprehensive study for implementing a machine learning technique (artificial neural network) to predict the thermal field evolution during the direct energy deposition of 316L stainless steel and tungsten carbides. The framework consists of a finite element thermal model and a neural network. The influence of the number of hidden layers and the number of nodes in each layer was also investigated. The results showed that an architecture based on 3 or 4 hidden layers and the rectified linear unit as the activation function lead to obtaining a high fidelity prediction with an accuracy exceeding 99%. The impact of the chosen architecture on the model accuracy and CPU usage was also highlighted. The proposed framework can be used to predict the thermal field when simulating multi-layer deposition.

Strategies for Improving Reliability of Solder Joints on Power Semiconductor Devices

2003 ◽  
Author(s):  
Guo-Quan Lu ◽  
Xingsheng Liu ◽  
Sihua Wen ◽  
Jesus Noel Calata ◽  
John G. Bai
Keyword(s):  
Finite Element ◽  
Solder Joint ◽  
Thermal Cycling ◽  
Flip Chip ◽  
Solder Joints ◽  
Mechanical Stresses ◽  
Flexible Substrates ◽  
Solder Joint Reliability ◽  
Joint Reliability ◽  
Reliability Of Solder Joints

There has been a significant research effort on area-array flip-chip solder joint technology in order to reduce package footprint, enhance current handling capability, and improve heat dissipation. However, there is a lingering concern over cyclic fatigue of solder alloys by thermo-mechanical stresses arising from mismatched thermal expansion coefficients of expansion among the various components of the package. In this paper, some strategies taken to improve the reliability of solder joints on power devices in single-device and multi-chip packages are presented. A strategy for improving solder joint reliability by adjusting solder joint geometry, underfilling and utilization of flexible substrates is discussed with emphasis on triple-stacked solder joints that resemble the shape of an hourglass. The hourglass shape relocates the highest inelastic strain away from the weaker interface with the chip to the bulk region of the joint while the underfill provides a load transfer from the joints. Flexible substrates can deform to relieve thermo-mechanical stresses. Thermal cycling data show significant improvements in reliability when these techniques are used. The design, testing, and finite-element analyses of an interconnection structure, termed the Dimple-Array Interconnect (DAI), for improving the solder joint reliability is also presented. In the DAI structure, a solder is used to join arrays of dimples pre-formed on a metal sheet onto the bonding pads of a device. Finite-element thermo-mechanical analyses and thermal cycling data show that the dimple-array solder joints are more fatigue-resistant than the conventional barrel-shaped solder joints in flip-chip IC packages.

A System for First Order Reliability Estimation of Solder Joints in Area Array Packages

1999 ◽  
Vol 122 (1) ◽  
pp. 6-12 ◽  
Author(s):  
Anand M. Deshpande ◽  
Ganesh Subbarayan ◽  
Dan Rose
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Solder Joints ◽  
Reliability Estimation ◽  
Ann Model ◽  
Analysis And Design ◽  
First Order ◽  
Simultaneous Analysis And Design ◽  
Reliability Of Solder Joints ◽  
Artificial Neural Network Ann

In this paper, a methodology, and a program based on the methodology, are presented, which, using the reliability of solder joints for various values of design/process parameters, can be estimated in seconds on a personal computer. The proposed methodology accurately captures the behavior of a solder joint in an artificial neural network (ANN) model trained to relate design as well as analysis parameters to fatigue life. The proposed methodology is novel since such simultaneous analysis and design (SAND) procedures do not appear to have been employed until now for reliability estimation of solder joints. The use of Moire´ interferometry as an experimental technique to estimate the analysis-related inputs to the neutral network model is also presented in the paper. [S1043-7398(00)01101-4]

A Hybrid Finite Element Modeling: Artificial Neural Network Approach for Predicting Solder Joint Fatigue Life in Wafer-Level Chip Scale Packages

2020 ◽  
Vol 143 (1) ◽  
Author(s):  
Zhiwen Chen ◽  
Zhao Zhang ◽  
Fang Dong ◽  
Sheng Liu ◽  
Li Liu
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Finite Element ◽  
Fatigue Life ◽  
Solder Joint ◽  
Finite Element Modeling ◽  
Life Prediction ◽  
Electronic Devices ◽  
Fatigue Life Prediction ◽  
Wafer Level

Abstract Fatigue life prediction of electronic devices is of great importance in both research and industry. Traditionally, fatigue tests and finite element modeling (FEM) are the two main methods. This paper presents a new hybrid approach (FEM combined with artificial neural network, (ANN)) for fatigue life prediction. Finite element models on wafer-level chip scale packages (WLCSP) with different chip thickness, PCB thickness, and solder joint pitches were created to evaluate the effect of structure parameters on the increase in maximum creep strain under thermal fatigue load. Modified Coffin–Manson equation was then employed to estimate the corresponding fatigue life. ANNs were built, and then trained, tested, and optimized with the datasets from modeling to predict increase in maximum creep strain and fatigue life. For the ANN built for strain prediction, prediction accuracy of the optimal network was 97% in accuracy tests and 93% in generalization tests. Accuracy of the other ANN for predicting fatigue life was 94.2% in accuracy tests and 88% in generalization tests. This hybrid method shows very promising application in fatigue life estimation of electronic devices which requires much less time and lower cost.

scholarly journals Prediction of Stress in Power Transformer Winding Conductors Using Artificial Neural Networks: Hyperparameter Analysis

Energies ◽  
2021 ◽  
Vol 14 (14) ◽  
pp. 4242
Author(s):  
Fausto Valencia ◽  
Hugo Arcos ◽  
Franklin Quilumba
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Finite Element ◽  
Power Transformer ◽  
Network Models ◽  
Activation Function ◽  
Neural Network Models ◽  
Transformer Winding ◽  
Artificial Neural ◽  
Femm Software

The purpose of this research is the evaluation of artificial neural network models in the prediction of stresses in a 400 MVA power transformer winding conductor caused by the circulation of fault currents. The models were compared considering the training, validation, and test data errors’ behavior. Different combinations of hyperparameters were analyzed based on the variation of architectures, optimizers, and activation functions. The data for the process was created from finite element simulations performed in the FEMM software. The design of the Artificial Neural Network was performed using the Keras framework. As a result, a model with one hidden layer was the best suited architecture for the problem at hand, with the optimizer Adam and the activation function ReLU. The final Artificial Neural Network model predictions were compared with the Finite Element Method results, showing good agreement but with a much shorter solution time.

Sign in / Sign up

Export Citation Format

Share Document