Windows Washing method of multiscale analysis of the in-situ nano-composites

We develop a fast and effective method of multiscale analysis of composite materials with a complex structure. The complexity of the structure results from the nano - TiC particles of the reinforcing phase being pushed into the interdendritic space during the manufacturing process. In such situations classic image analysis methods are ineffective because they are not able to resolve the high heterogeneity of structure and problems with phase separation. Moreover, even if the image analysis methods are successful, their application to calculate the macroscopic properties is highly problematic because it requires massive data processing. The developed Windows Washing methodology is a multiscale procedure which combines image analysis, statistics, Fourier analysis and analytical Representative Volume Element methods. From the mathematical standpoint we are interested in direct approximate computations of the basic, structural sums. All necessary input concerning the structure of materials is introduced from the real experimental data. There are some inherent limitations on experimental accuracy of determining the positions of particles. But they could be overcome by the complex methodology of Windows Washing. The more efforts we apply the more accurate reconstruction of the positions could be achieved. In fact, instead of real particles we use their images quantified by pixels. The qualitative relationship between the structural image and macro-properties of the nano -TiC composite is found directly on the 2D structural image's pixel level. The developed method establishes a solid foundation for building a database for classification of the composite structures. The appropriate architecture of data analysis, called Extract, Transform, and Load is included in the developed method.

Automated summary

We have developed a fast and effective method for multiscale analysis of complex structured composite materials. This method overcomes the limitations of classic image analysis methods in resolving heterogeneity and phase separation issues. It combines image analysis, statistics, Fourier analysis, and analytical Representative Volume Element methods to compute structural sums and calculate macroscopic properties. The method is based on real experimental data and allows for accurate reconstruction of particle positions. It establishes a solid foundation for building a database for classifying composite structures.