Life Prediction Model

The PCM method is also proved to be suitable for safe operation under stressed battery operating conditions. 

Due to the lack of adequate knowledge of the specific degradation mechanisms causing the deviation between the expected average performance and the measured cell performance, the present fife estimation model relies on a simplified empirical degradation model with a few parameters.Thomas et al. [82] described an empirical degradation model with a single stress factor, temperature, for the expected relative resistance, fi T t)  

The error model accounts for variabihty due to measurement error as well as intrinsic ceU-to-ceU differences. Thomas etal. [82] recommended the following error model, which combines the measurement error and the unique behavior of the ith individual ceU. 

The parameters associated with the degradation and error models are estimated using linear regression. Finally, the fitted degradation model can be used to estimate the mean lifetime of the ceU at a specified temperature for a given end-of-life criterion. The details of this battery life estimation methodology along with Monte Carlo simulation to assess lack-of-fit statistic are provided in Ref. [82]. 

---

Objective of monitoring. A monitoring system, eventually with computerised data acquisition, should meet specifically defined objectives, such as a) to monitor the durability of the structure and its condition in order to make timely decisions for preventive and/or repair actions, b) to monitor the effect of preventative or repair actions, c) to monitor the condition of stmctures based on new materials and/or new technology (including service-life prediction models), d) to follow the time development in areas where access is difficult. 

For these reasons, we decided to consider these criteria with the aim of assessing their reliability in terms of life estimation, by comparing their predictions with some of the experimental results taken from the extensive database available. This would be of help in obtaining information useful for design purposes, like strengths and weaknesses of each criterion, and, at the same time, in further clarifying the directions and the need for the development of life prediction models of general applicability. 

Without proper knowledge of the circumstances in which degradation mechanisms are active and of how they interact, there is no firm base for reliable life prediction models. Products will be over-designed to compensate for the lack of accurate predictions. The models presently available for quantifying the degradation and ageing mechanisms presented in the previous section are reviewed here. 

Wang, W. and W. Zhang (2008, July). An asset residual life prediction model based on expert judgments. European Journal of Operational Research 188(2), 496-505. 

Confidence intervals are essential for component strength and life prediction methods, and for methods verification in this program. Verification of the life prediction methods will be accomplished by comparing observed confirmatory specimen lives with predictions. There will be some uncertainty in the predictions, due to the size and number of specimens tested to generate the life prediction model parameters. Confidence intervals on the predictions will help quantify this uncertainty, and thereby determine (1) the expected deviation between measured and calculated lives, or (2) if the deviation is a result of modeling inaccuracies. Confidence intervals are also needed for component design to define the lower limits of reliable component operations. 

Weibull life distribution model is selected which has previously been used successfully for the same or a similar failure mechanism. The Weibull distribution is used to find the reliability of the life data and it helps in selecting the particular data that is to be used in life prediction model. The Weibull distribution uses two parameters, namely, b and to estimate the reliability of the life data, b is referred to as shape parameter and is referred to as scale parameter. 

Some of the fatigue-life prediction models that have been employed for lead-free solder joints are outlined in Table 59.7. In addition, other models for predicting fatigue life from the finite element models and performing relative comparative analysis can be foimd in References 16 and 47. 

Syed, Ahmer, Accumulated Creep Strain and Energy Density Based Thermal Fatigue Life Prediction Models for SnAgCu Solder Joints, Electronic Components and Technology Conference, 2004, pp. 737-746. Corrected version available online at http //www.amkor. com/products/notes papers/asyed ectc2004 corrected.pdf. 

As noted in previous sections, the development of life prediction models for the reliability of patterned features such as periodic lines on substrates inevitably requires knowledge of intrinsic stress and mismatch stress generated during film growth, patterning, passivation and service. In this section, three prominent experimental methods for determining stress in thin films with patterned lines are considered the substrate curvature method, the x-ray diffraction method, and the micro-Raman spectroscopic method. The advantages and limitations of each of these techniques are also briefly addressed. 

The kinetics of environmental attack will be governed by the rate-determining step in the overall mechanism of failure and very little information exists on such details. Indeed, the lack of such information has been a severe handicap in developing life prediction models, where clearly a knowledge of the kinetics is a crucial aspect which is needed in many types of time-dependent models. 

The above studies showed that, for this particular adhesive system, the rate of diffusion of water through the adhesive to the interface was the rate controlling step. Now, if the diffusion of water through the adhesive is Fickian in nature, then the concentration profile of water as a function of time into the joint may be calculated [6,43,44], and such information may then be readily used in life-prediction models, as discussed below. It should be noted that the values of the diffusion coefficient given in Fig. 12b, which are very typical for structural epoxy adhesives, lead to the conclusion that, at ambient temperatures, it will take at least a year or more for the adhesive layer in a joint, say about 20 mm x 10 mm in size (as often used in single-overlap shear joints), to reach its uniform, equilibrium concentration of water, although of course, depending upon the details of the adhesive system , even complete failure of the joint due to environmental attack may have occurred well before this time is reached. 

The overall approach in developing the life-prediction model is [73,74] as follows. 

Traditional PHM extracts performance degradation pattern from real-time monitoring signal data or acquire degradation parameter data directly. Based on that information, life prediction model can be given according to the nature of data or the generalization capability of the selected models to... 

Model How to establish reasonable life prediction models and simultaneously verify them for new products ... 

With the main line of life cycle stages, we introduce the interdependent relationship of the data and model under this framework, and provide their support for life prediction modelling. 

This process uses historical and similar product information, and expert knowledge as inputs. The output is the theoretical life prediction model set Ml (Fig. 2). 

Life prediction modelling for Double nozzle flapper electro-hydraulic servo valve... 

For the purpose of evaluating the life and reliability indexes of servo valve, both simulation test and accelerated test are conducted before that is put into use. The information obtained from the two tests are used for life prediction modelling according to the procedure in Section 3.1 and 3.2. The results are shown in Figure 5. At first, the wear of nozzle flapper and sliding valve under the influence of oil pollution is simulated through... 

Life prediction modelling and on-line updating for super luminescent diode... 

Life Prediction Models. The reliability test results discussed in this chapter are under accelerated conditions. For the data to be of use in gauging the reliability of actual product boards, accelerated test results need to be extrapolated to field conditions. Without acceleration factors and/or calibrated life prediction models of board level reliability under representative field conditions, product reliability claims remain unfounded. In the absence of such a bridge from test to field conditions, real products can be over-designed, at risk, or even rejected under assumptions with variable degrees of conserva-... 

---