+ All Categories
Home > Documents > Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant...

Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant...

Date post: 22-Dec-2015
Category:
Upload: beatrice-miranda-tate
View: 213 times
Download: 0 times
Share this document with a friend
Popular Tags:
15
Warranty Forecasting of Electronic Boards using Short-term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University www.ecc.itu.edu.tr
Transcript
Page 1: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Warranty Forecasting of Electronic Boards using Short-

term Field Data

Mustafa Altun, PhDAssistant Professor

Istanbul Technical Universitywww.ecc.itu.edu.tr

Page 2: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Outline

MOTIVATION Overview of reliability prediction techniques. Unexpected failures in the field.

GOAL Warranty forecasting using short-term data.

METHOD Modeling for an old board.

Filtering the data and developing the model. Prediction for a new board.

Estimation methods: maximum likelihood, Bayesian, rank regression.

RESULTS

Page 3: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Outline

MOTIVATION Overview of reliability prediction techniques. Unexpected failures in the field.

GOAL Warranty forecasting using short-term data.

METHOD Modeling for an old board.

Filtering the data and developing the model. Prediction for a new board.

Estimation methods: maximum likelihood, Bayesian, rank regression.

RESULTS

Page 4: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Overview of Reliability Prediction

Field Data

Analysis

Accelerated Tests

Simulation with PoF

Pass-Fail Tests

ACCURACY

TEST DURATION

REAL-TIME PERFORMANCE

PREDICTION BEFORE FIELD

COST

Red cells for the worstBlue cells for the bestWhite cells for the average

Page 5: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Unexpected Failures in the Field.

Unexpected failure rates in «Early Failure». How to predict the future?

What is the proper amount of time that the product should stay in the field?

Page 6: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Prediction with Field Return Data

Conventioanlly, «useful life» or «wear out» period can not be predicted using the data from «early failure».

We overcome this problem!

Goal: predicting reliability of electronic boards throughout their 3-year warranty with 3-month field data.

Page 7: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Outline

MOTIVATION Overview of reliability prediction techniques. Unexpected failures in the field.

GOAL Warranty forecasting using short-term data.

METHOD Modeling for an old board.

Filtering the data and developing the model. Prediction for a new board.

Estimation methods: maximum likelihood, Bayesian, rank regression.

RESULTS

Page 8: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Modeling for an Old Board Use full field data for an old electronic board. Filter the data to eliminate incomplete and poorly collected data.

Based on Weibull β parameter. Targets on hidden errors

Model the filtered data with Weibull distribution.β values for different assembly times vs. assembly time intervals

Page 9: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Modeling for Weibull β Parameter

Modeling for β: 1. Creating a model .

product dependent parameter technology dependent parameter

Page 10: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Prediction for a New Board

Prediction method: Bayesian fitting for Weibull (prior exponential) β parameter

MLE vs. Bayesian Weibull distribution is used Bayesian overwhelms MLE for small samples.

Page 11: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Prediction for a New Board

Reliability prediction for β: 1. Creating a model2. Using the model for prediction of a new board.

Use same . Determine using 3-month data of the new board.

Page 12: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Experimental Results

Experimental Results

Page 13: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Experimental Results

Experimental Results

Board-K is not a member of the family of Board B-E-F.

Page 14: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Conclusions – Future Work

We make a 3-year warranty forecasting of a new electronic board having its 3-month field data.

We develop a mathematical model of β as a function of field data time interval using board dependent parameters.

The predicted results from our method and the direct results from the field return data matches well.

We will improve our model, more generic, to be applicable for different electronic products.

Page 15: Warranty Forecasting of Electronic Boards using Short- term Field Data Mustafa Altun, PhD Assistant Professor Istanbul Technical University .

Thank you!


Recommended