Test Data Processing Method of Fracture Experiments of Dam Concrete for Inverse Analysis

In recent years, researchers identify the complete softening curve of concrete materials by employing inverse analysis method. The inverse analysis method based on fracture experiments of concrete is numerical simulations by using discrete approach of non-linear fracture mechanics. First, a softening curve is assumed and the load-displacement curve obtained by the numerical simulation is compared to the one determined experimentally by fracture experiment of concrete. By updating the assumed softening curve, the numerical results optimally fit the measured ones. When the best fit of the numerical and the experimental results is obtained, the assumed softening curve is viewed as the one for characterization of the material behavior. Load-crack mouth opening displacement curve (P-CMOD curve) is an important measured curve which characterizes fracture property of concrete in fracture experiment. For each group of fracture experiment specimens, there are three companion specimens tested in the group. The emphasis of this paper is as follows: Developed a processing program for the original test data which length is up to 30000~300000 lines for each specimen. The program can filter test data scattered far from the P-CMOD curve and remove the effect of fluctuation in the measurement of fracture experiment. A new test data processing method for fracture experiment of concrete was proposed based on crack propagation process of concrete materials. Thus a representative P-CMOD curve which characterizes the fracture property of the companion specimens for each concrete specimen group is achieved by averaging the above-mentioned test results of three companion specimens. Since the number of data points used in optimal fitting of inverse analysis have an effect on the computer running time observably, the minimum number of data points adequately representing P-CMOD curve is desirable. The minimum of data points were extracted from the averaged data for the companion specimens by an optimization procedure. This paper employed the proposed method to analyze the original test data of different sizes of wedge-splitting specimens made of dam concrete. The representative P-CMOD curves were obtained for companion specimens of different fracture specimen groups. The obtained representative P-CMOD curves were prepared for an inverse analysis to find the softening curves of dam concrete and were employed to investigate the fracture property of dam concrete.