Avaliação de distribuições de probabilidade no estudo da resistência características de blocos de concreto

Abstract

This comprehensive study meticulously evaluated six probability distribution models for the compressive strength of 24 concrete blocks, 12 with metakaolin and 12 without addition. The models analyzed were Lognormal, Gamma, Normal, Weibull, Beta, and Exponential. For blocks with metakaolin, the Lognormal distribution presented the best fit, followed by Gamma, Normal, Beta, and Weibull, while the Exponential needed to be more adequate. In blocks without metakaolin, the Gamma distribution was the most efficient, followed by Normal, Weibull, Lognormal, and Beta, with the Exponential also inadequate. The model parameters were adjusted based on the resistance frequencies estimated for each group of samples. Thus, Lognormal and Gamma performed better in predicting compressive strength, depending on the presence or absence of metakaolin. The Exponential distribution, in turn, was considered inadequate in both cases.