Incorporating invalid test results in glass fracture strength determination

Area of Science:

• Materials Science
• Structural Engineering
• Statistical Modeling

Background:

• Standardized tests for glass fracture strength often discard invalid results.
• Invalid tests, where fracture doesn't initiate in the inner ring, contain valuable data.
• Neglecting invalid test data can lead to biased estimations of glass strength.

Purpose of the Study:

• To propose a Bayesian inference approach for incorporating invalid test data.
• To assess the impact of invalid tests on glass strength characterization and structural reliability.
• To improve the efficiency and accuracy of glass strength predictions.

Main Methods:

• Application of Bayesian inference to existing glass fracture strength datasets.
• Analysis of data at ambient and elevated temperatures.
• Quantification of the influence of test number and invalid test inclusion on strength estimation and reliability.

Main Results:

• Discarding invalid tests results in biased glass strength estimates.
• Incorporating invalid tests improves prediction precision, potentially reducing the required number of tests.
• Ignoring invalid tests leads to an overestimation of the structural reliability index for glass slabs.

Conclusions:

• Invalid test data should be incorporated for more accurate glass strength characterization.
• The proposed Bayesian method enhances testing efficiency and refines strength predictions.
• Considering invalid tests is crucial for robust reliability evaluations of glass structures.