Concrete is commonly used in protective and civil structures, like tunnels and storage buildings. For the design and reliable safety assessment of such structures it is very important to know the behaviour of concrete under static loading conditions as well as under impulsive loading. To understand the behaviour of concrete under impulsive loading and to be able to predict the failure behaviour, it is important to know the influence of the loading rate on concrete strength and failure parameters and to quantify how they change with increasing loading rate. The fact that material parameters depend on the applied loading rate is called rate dependency. Objectives The mechanical response of concrete structures is often predicted with numerical material models in a finite element context. To properly predict the response of structures under impulsive loading, the rate dependency should be included explicitly in the material model. To validate a physically realistic concrete material model for high loading rates, experimental data on the rate dependency of concrete tensile strength and fracture properties is needed. Data which shows the influence of the loading rate on the mechanical parameters like strength and stiffness can be found in literature for low as well as high loading rates (> 1000 GPa/s). However, data on the rate dependency of the fracture energy, and especially the fracture behaviour, is scarce. Therefore, the objective of this study is to quantify the rate effect on the tensile strength and stiffness at medium and high loading rates, as well as to determine the stress-displacement relation that reflects the fracture process and determines the fracture energy. Besides experimental data on the rate dependency of concrete properties, it is also important to understand the physical mechanisms behind the rate effects. The physical mechanisms will explain the change in concrete behaviour under dynamic loading conditions. A detailed analysis of the causes behind the rate dependency of concrete is missing in literature. There is no common explanation of the underlying mechanisms of the rate dependency of concrete tensile properties. Therefore, one of the objectives of this study is to identify the different mechanisms behind the rate effects on tensile properties and to quantify the influence of the different mechanisms. To gain detailed information on the mechanisms behind the rate dependency and to be able to quantify the influence of the different mechanisms, information on the rate effects on fracture characteristics (width of the fracture zone, crack distribution and crack /fracture lengths) is needed. The experimental research (at macro level) is, therefore, combined with microscopic research. Experimental program A research program has been developed to study the rate dependency of concrete tensile fracture properties. To study the rate effect on the tensile properties of concrete, uniaxial tensile test are conducted at three different loading rates: - Static loading rate as a reference; loading rate 10-4 GPa/s; - Moderate loading rate with the gravity driven Split Hopkinson Bar set-up (SHB) at the Delft University of Technology; loading rate 50 GPa/s; - High loading rate with a newly developed Modified Split Hopkinson Bar set-up (MSHB) at the laboratory of TNO in Rijswijk; loading rate > 1000 GPa/s. From literature it was concluded that the moisture in the pores plays an important role in the rate dependency of concrete properties, especially in the moderate loading rate regime (up to 50 GPa/s). Therefore, it was decided to study the influence of moisture on the rate dependency of the tensile fracture properties by varying the loading rate, moisture content and micro structure. To study the influence of the micro structure, two types of concretes are used; Portland cement concrete and Blast Furnace Slag (BFS) cement concrete. BFS cement concrete has a denser micro structure with less capillary pores. To study the influence of moisture in the pore system on the rate effects on concrete tensile properties, the concrete specimens have been subjected to four different moisture conditions for approximately 21 days (after being drilled out of the cubes at an age of 28 days). The four different moisture conditions are: - “Normal” condition: specimens are stored under controlled conditions of 20°C and 50% RH; - “Wet” condition: specimens are immersed in water; - “Dry-50” condition: specimens are dried in an oven of 50°C and 15% RH; - “Dry-105” condition: specimens are dried in an oven of 105°C and 2% RH. After the uniaxial tensile tests are finished, the fracture patterns have been studied by impregnating the cracks with epoxy and studying them by microscope. The influence of the rate dependency on the fracture parameters has been determined by quantifying the rate effects on the lengths of the different cracks and on the width of the fracture zone. Test set-ups For the static tests (loading rate 10-4 GPa/s), deformation controlled uniaxial tensile tests have been performed on cylindrical concrete specimens (Ø 74 mm, length 100 mm). The gravity driven Split Hopkinson Bar is used to conduct the uniaxial tensile tests at moderate loading rate (50 GPa/s). This SHB set-up consists of two cylindrical aluminium bars between which the concrete specimen (Ø 74 mm, length 100 mm) is glued. The tensile stress wave is generated with a drop weight, which slides along the lower bar and hits an anvil at the bottom end. The tensile wave travels upwards through the aluminium bar and through the specimen, fracturing the specimen when the tensile strength of the concrete is reached. For the very high loading rates (> 1000 GPa/s) a new Modified Split Hopkinson Bar (MSHB) set-up is used. The Modified Split Hopkinson Bar is based on a different principle than the Split Hopkinson Bar, i.e. the principle of spalling. A shock wave is introduced into a horizontal steel bar by detonating an explosive charge at one end of the bar. At the other end, the concrete specimen is attached which is first loaded in compression and will fail in tension due to the reflected tensile wave (spalling). In the Modified Split Hopkinson Bar set-up a new innovative measurement technique was used, enabling direct measurement of strains and deformations of the loaded specimens. The direct measurement method used in the Split Hopkinson Bar and Modified Split Hopkinson Bar set-ups generates information on real-time strains and deformations. This makes it possible to reconstruct failure behaviour and has given insight in the different causes of the observed rate effects. Experimental results For the moderate loading rates (50 GPa/s) an increase of the tensile strength of approximately 2 MPa is found for dry and normally cured concrete compared to the static tests. For wet concrete, the increase in tensile strength was found to be more pronounced, approximately 4 MPa for Portland cement concrete and 3 MPa for Blast Furnace Slag cement concrete. At high loading rates (1700 – 2450 GPa/s), the tensile strength results have shown an increase of approximately 5-7 MPa for normal and dry concrete and 12-15 MPa for wet concrete. The fracture characteristics, i.e. fracture lengths and widths of the fracture zone, and the failure behaviour have been quantified by measuring the individual and summarized crack lengths and determining the stress-deformation curves. From the results it was concluded that the width and length of the macro fracture are not influenced by the loading rate. For normally cured and dry concrete and loading rates up to 50 GPa/s, the fracture energy, shape of the stress-deformation curve, width of the fracture zone and the number of micro cracks are hardly affected by the loading rate. Wet concrete shows enhanced resistance in the post-peak phase of the stress-deformation curve. For the high loading rate regime (> 50 GPa/s) the total summarized length of the micro cracks and the width of the fracture zone increase considerably, as well as the resistance in the post-peak phase of the stressdeformation curve and the fracture energy. Mechanisms With the available experimental data on tensile strength, fracture characteristics and postpeak failure behaviour, the failure mechanisms have been reconstructed and the main causes for rate dependency of the tensile strength and fracture resistance have been identified. The most important possible causes are (1) structural inertia of the fracture zone, (2) influence of inertia at micro-level which can delay crack initiation and propagation, (3) additional micro cracking and (4) enhanced fracture resistance caused by moisture in the pores (Stefan effect). By using basic principles of fracture mechanics and a simple model based on the Stefan effect, the different mechanisms and the loading rates at which these mechanisms have significant effect have been determined. Nuclear Magnetic Resonance and Mercury Intrusion Porosimetry results provide necessary data on moisture distribution in the pore system. Structural inertia effects For the experimental data from the presented research and the method used to measure and analyse the data (1D approach), it was shown that structural inertia of the fracture zone (axial direction) does not contribute to the obtained enhanced tensile strength for the moderate as well as the high loading rate regime. Also, structural inertia of the fracture zone hardly affects the post peak behaviour of concrete. This holds for all concrete types and moisture contents studied. Effect of micro inertia, Stefan effect and additional cracking The mechanism which causes the strength increase due to moisture in the pores is the so called Stefan effect. The Stefan effect is explained as the reaction force, which is induced when two plates with moisture in between are separated. The equation to calculate the Stefan effect is modified to fit concrete pore structure properties. To incorporate the concrete structure, it is assumed that the Stefan effect is valid for cylindrical pores. From the NMR data combined with the experimental data on strength and fracture energy and the Stefan effect model, it was concluded that only the moisture in the capillary pores contributes to the enhanced tensile strength and post-peak resistance. The dominant mechanism in the moderate loading rate regime (< 50 GPa/s) causing the tensile strength increase due to moisture in the pores is the Stefan effect. In the high loading rate regime, both the Stefan effect and the micro inertia effects on crack propagation due to the limitation on crack velocity contribute to the increase in tensile strength. The enhanced fracture resistance in the post peak behavior for wet concrete and moderate loading rates up to 50 GPa/s has been ascribed to the viscous effects of moisture in the concrete pores (Stefan effect). For the high loading rate regime, loading rates exceeding 50 GPa/s, the enhanced resistance in the post-peak behaviour is partly caused by the formation of additional (micro) cracks. However, the increase in (micro) crack length is insufficient to explain the observed high fracture energy increase for the high loading rate regime. Therefore, the increased fracture toughness in the post-peak behaviour in the high loading rate regime has been explained by the formation of additional (micro) cracks as well as the viscous behaviour of concrete (wet concrete) and by the micro inertia effects due to material inherent limitation of the crack velocity. Future application The acquired knowledge on the mechanisms behind the rate dependency of concrete fracture properties can be used to improve numerical models. One of the most commonly used approaches to model concrete fracture is the application of continuum models. The properties and behaviour of the elements in a continuum model are defined by a constitutive law, which also determines the change in properties when fracture takes place. For modelling failure under dynamic loading conditions, the mechanisms behind the rate effects on concrete tensile properties, i.e. moisture in the capillary pores, additional micro cracking and micro inertia effects due to limitations on crack velocity, should be incorporated into the constitutive material model. Which mechanisms should be incorporated into the material model depends on the scale at which modelling takes place. The presented research has also shown that for modelling dynamic experiments the internal material length scale, an important parameter to model fracture processes, should be based on the width of the macro fracture zone and connected micro cracks. This zone contains the macro crack, which physically separates the specimen in two halves, as well as the micro cracks that are attached/connected to the macro crack. The width of the macro fracture zone with connected micro cracks does not significantly change with increasing loading rate. Subsequently, the internal length scale as defined for a specific model also does not significantly change with increasing loading rate. The dominant mechanisms found in the presented research can be implemented in dynamic models and the acquired data set1 can be used to validate the developed models. With the suggested dominant mechanisms and knowledge on the rate dependency of the concrete tensile strength and fracture behaviour, the response of concrete materials under dynamic loading can now be better understood and predicted more accurately.