Application of an Empirical Extreme Value Distribution to Load Models

Area of Science:

• Extreme value theory
• Probability distributions
• Statistical modeling

Background:

• Accurate modeling of extreme loads is crucial for structural safety and risk assessment.
• Existing extreme value distributions may have limitations in representing real-world load data.
• The need for robust probability distribution models for various extreme intensities.

Purpose of the Study:

• To propose and evaluate a novel empirical extreme value distribution with defined lower and upper bounds.
• To apply the proposed distribution to model maximum load intensities from earthquake ground motion, wind speed, and supermarket live loads.
• To assess the performance of the proposed distribution against existing models.

Main Methods:

• Development of an empirical extreme value distribution incorporating lower and upper bounds.
• Application of the distribution to datasets of annual maximum earthquake ground motion, wind speed, and supermarket live loads.
• Parameter estimation and comparison with established probability distribution models.

Main Results:

• The proposed distribution demonstrates considerable improvements over other models for annual maximum earthquake ground motion.
• Challenges in parameter estimation, particularly determining the upper bound value, were noted.
• The distribution shows promise for modeling wind speed and live load extremes, with potential for further refinement.

Conclusions:

• The novel empirical extreme value distribution offers a valuable tool for analyzing extreme load intensities.
• The model shows particular efficacy in representing earthquake ground motion extremes.
• Further research is warranted to optimize parameter estimation and explore broader applications for wind and live load modeling.