ICF10B

Figure 3: Bilinear relation for strain softening. TABLE 2: COMPRESSIVE STRENGTH AND ELASTIC MODULUS AT AN AGE OF 14 DAYS Mix A Mix B Mix C Compressive strength N/mm2 35,3 57,4 112,2 Modulus of elasticity N/ mm2 38.000 45.000 49.000 Fracture mechanics parameters have been determined by means of the wedge splitting test [7]. These tests have been carried out on cubes with a sawn notch having half the heigth of the cube. Results obtained after inverse analysis are shown in Fig. 3. The insert in Fig. 3 summarizes the obtained parameters of the bilinear model. In the last line the characteristic length lc is given as calculated by the following equation: lc = EGf ft 2 (2) This length can be considered as an indication of the brittleness of a material. In addition to the mechanical properties we also need the hygral diffusion coefficient and the shrinkage as function of relative humidity. The hygral diffusion coefficient can be calculated from measured drying data [8]. On this basis it is possible to calculate time-dependent moisture distributions in concrete elements prepared with different types of concrete and under different boundary conditions in order to predict crack formation. 3. NUMERICAL SIMULATION OF CRACK FORMATION In the preceeding section we have outlined all the material parameters which have to be known or determined before the risk of crack formation can be realistically estimated. As a next step we will simulate numerically crack formation in a composite structure such as concrete. We consider concrete to be built up by coarse aggregates embedded in the mortar matrix. As we want to compare crack propagation in ordinary and high strength concrete we need parameters of two matrixes, i.e. a cement-based mortar with a w/c-ratio of about 0,5 and 0,35 respectively. Based on earlier experiments the data compiled in Table 3 have been used for the numerical simulation. To simulate crack propagation a specimen used for wedge splitting experiments has been chosen. The composite structure of concrete has been simulated by generating randomly dispersed hexagonal

RkJQdWJsaXNoZXIy MjM0NDE=