The failure strength of concrete materials has been widely shown to be dependent on experimental parameters such as specimen geometry and strain-rate. The effects of specimen geometry have been shown both theoretically and experimentally to be a result of the quasi-brittle nature of concrete. While the failure strength of concrete has been widely reported to increase significantly when deformed at high strain-rates, the physical mechanisms driving this phenomenon remain the source of debate amongst researchers. This means that constitutive models designed to predict this rate dependent behavior are not based on the real physical mechanisms that drive this behavior but rather fit to a limited set of experimental results. This leads to numerical models that often do not scale accurately across different geometries or stress states. A better understanding of the dynamic failure behavior of quasi-brittle materials can be used to develop more robust multi-scale material models capable of predicting the response and failure of large-scale structures under impact loadings

The Kolsky (Split Hopkinson) Bar is a well-established experimental technique capable of applying controlled and repeatable dynamic loading conditions to a specimen or structure and measuring the resulting mechanical response with high temporal resolution. Kolsky bar based experimental techniques are used throughout this thesis to characterize the dynamic behavior of high strength concretes, including the effects of both strain rate and specimen size. It is shown that the failure strength of the materials studied is highly dependent on both specimen size and strain rate. However, the ratio of the failure strengths measured at quasi-static and dynamic strain-rates, a parameter known as the Dynamic Increase Factor (DIF), exhibits no clear dependence on specimen size but is only a function of strain rate. Additionally, it is shown that through careful and proper experimental design, experimental errors, which are often touted as the source of the observed rate dependence, can be significantly minimized. While this reduction in experimental error results in a DIF that is less than historical data, a significant increase in failure strength is still observed. This suggests that the rate dependence is representative of the real specimen response. Gaining deeper insights into the physical mechanisms that drive this rate dependence requires the development of experimental techniques that facilitate the detailed observation of specimen behavior as a result of applied dynamic loadings.

The short duration of typical dynamic loading experiments limits the capabilities of in-situ observation techniques. While techniques such as high-speed photography are often used for in-situ observation of dynamic experiments, the resulting observations and measurements are typically two dimensional. These techniques are therefore incapable of providing information beyond limited qualitative analysis of the inherently three-dimensional failure process in heterogeneous materials such as concrete. Alternative observation techniques, such as X-ray computed microtomography (micro-CT) provide high-resolution three-dimensional imaging of the specimen interior, however this capability comes at the sacrifice of temporal resolution.

In this dissertation, the highly controlled dynamic loading capabilities of Kolsky Bar based experimental techniques are coupled with the high-resolution three-dimensional imaging of X-ray micro-CT to gain deeper insights into the failure behavior of concrete specimens under dynamic loading conditions. This is accomplished by designing dynamic loading conditions such that the failure process in concrete specimens is initiated but the load is removed before complete failure occurs. This facilitates X-ray micro-CT analysis of the recovered specimens to investigate changes in the internal specimen morphology as a result of the applied dynamic loading, including the initiation of crack propagation. However, identifying loading induced cracks in tomographic images of porous solids presents its own challenges. These challenges are overcome in this dissertation by developing a geometric segmentation algorithm that identifies cracks in porous solids based on the significantly different geometries formed by cracks as compared to naturally occurring voids.

This novel experimental technique is used to study the failure behavior of high-strength concrete specimens subjected to short-duration, constant-stress loadings. It is shown that these specimens are capable of supporting an applied load above their quasi-static failure strength for short durations. This duration is found to be dependent on the applied stress, with the duration increasing exponentially as the applied overload approaches the quasi-static failure strength. Additionally, the morphological evolution of a single concrete specimen subjected to six constant stress loading cycles above its quasi-static failure strength is studied using X-ray micro-CT. These results revealed that the failure process is comprised of a period of void compaction, likely accompanied by the nucleation and growth of microcracks, followed by the coalescence and propagation of a major crack network that leads to complete specimen collapse.

Finally, this experimental technique is applied to study the behavior of concrete specimens subjected to high strain rate loading. This is accomplished using a modified Kolsky compression bar with momentum trapping capability to achieve dynamic constant strain-rate deformation before rapid unloading at a pre-determined stress level, allowing specimens to be recovered and scanned. A statistical analysis of the data collected from experiments of this type is then performed to estimate the stress level at which strain localization and major crack propagation occurs at high strain-rates. Results from this novel experimental technique reveal that, a significant portion of the total dynamic strength increase observed in the materials studied occurs after the onset of major crack propagation. This indicates that a significant portion of the dynamic strength increase occurs during the fragmentation process and should not be considered representative of the intrinsic material properties.

The novel experimental techniques developed in this dissertation provide a clearer picture of the behavior of concrete specimens under dynamic loading conditions. The insights gained using these techniques can be used to inform the development of more accurate constitutive models based on the real physical mechanisms driving the observed material behavior.

Degree Date

Fall 12-21-2019

Document Type


Degree Name



Mechanical Engineering


Xu Nie

Subject Area

Mechanical Engineering

Number of Pages




Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial 4.0 License