In general, there are large differences between shrinkage strains predicted by the existing Models (ACT Model, BP Model).
Thus, the statistical method is used in order to improve the predictions of shrinkage strains.
The statistical method is able to use for the good long-time predictions of shrinkage strains from short-time measured data by using latin hypercube sampling of random parameters and Bayesian theory that utilizes the shrinkage strains by existing Models as the prior.
At this time, the closer Models are to short-time measured data, the more agreeable the predicted values are to actual values.
In this paper, BP Model is found suitable for Bayesian extrapolation, and predicts long-time shrinkage strains better than ACI Model does.
