On Stochastic Approach to Evaluation of Service Life for Reinforced Concrete Constructional Elements and Structures during Concrete Carbonization

The currently existing deterministic models for determination of structure service life do not take into account to the full extent variety of factors that have an effect on them and also stochastic characteristics of the used natural materials and conditions for manufacturing construction products, possible errors during the process of designing and construction etc. More trustworthy approach determining longevity and growth rate of destruction processes for structures in future periods can be obtained with the help of statistic methods that take into account a probabilistic essence of the process. The paper presents a possible approach of the probabilistic analysis on reinforced concrete structure service life while assessing rates of variation in depth growth of carbonization in a concrete protective layer on the basis of the existing experimental and calculated data pertaining to changes of the given index in reinforced concrete structures of various types. Variability of the existing approaches for determination of structure longevity as a whole has been shown firstly due to various number of basic vаriables used in calculations. Stochastic processing of the data on parameters of carbonization depth in the concrete protective layer has been carried out and this processing has made it possible to determine variation rates which allow to assess the presupposed service life of reinforced concrete structures having similar characteristics and being operated under analogous conditions. A definitive non-uniformity in statistic indices has been established that testifies about the necessity to increase accumulation of data on the investigated characteristics and to execute its processing more thoroughly. An expedience in usage of a concrete impermeability as a main factor determining its longevity has been established on the basis of statistical assessment of the existing data.