Prediction of Cement-Based Materials' Water Content with the Use of Electromagnetic Homogenisation Schemes

Concrete delayed behaviour such as creep or shrinkage is closely linked to the moisture transfers within it. Nondestructive methods used to determine the water content in civil engineering structures are often based on electromagnetic waves propagation through the cementitious medium (RADAR, Time or Frequency Domain Reflectometry, capacitive measurements). However, it is still difficult to connect the measured macroscopic permittivity to pores' saturation degree as the measured property does not only depend on the material's water content. Geometrical arrangement of the structural components as well as the property of each of them can influence the effective permittivity value. The study presents the application of electromagnetic homogenisation schemes like the Symmetric Bruggeman model to connect measured permittivities with materials saturation. These models enable to take into account microstructure specificity depending on the material under test. In the study, a large open-ended coaxial probe was designed to investigate the permittivity of concrete as well as its phases and components properties. The large diameter of such probe enables to investigate large volumes within the material under test. Measuring the reflection coefficient with the use of a Vector Network Analyser and applying a capacitive model to the collected data enables to determine the samples' permittivity. A homogenisation model taking into account the complex microstructure of cements paste and concrete is then built. Permittivity of the phases are used as an entry for the simulation. Analytically derived results are compared with measurements performed onto cement bases materials with the large open-ended coaxial. Comparison between simulation and experiments shows that a combination of homogenisation processes at the different scales of concrete enables a great improvement in water content prediction compared to classical empirical or calibration laws.