Hanley Kj

Fabric and effective stress distribution in internally unstable soils

Internal instability is a form of internal erosion in broadly graded cohesionless soils in which fine particles can be eroded at lower hydraulic gradients than predicted by classical theory for piping or heave. A key mechanism enabling internal instability is the formation of a stress-transmitting matrix dominated by the coarse particles, which leaves the finer particles under lower effective stress. In this study, discrete element modeling is used to analyze the fabric and effective stress distribution within idealized gap-graded samples with varying potential for internal stability. The reduction in stress within the finer fraction of the materials is directly quantified from grain-scale data. The particle-size distribution, percentage finer fraction, and relative density are found to influence the stress distribution. In particular, effective stress transfer within a critical finer fraction between 24 and 35% is shown to be highly sensitive to relative density.

EFFICIENT CALIBRATION OF DISCRETE ELEMENT MATERIAL MODEL PARAMETERS USING LATIN HYPERCUBE SAMPLING AND KRIGING

Material model parameter identification for discrete element models (DEM) is typically done using a trial-and-error approach and its outcome depends largely on the experience of the DEM user. This paper describes a work flow which facilitates the efficient and systematic calibration of discrete element material models against experimental data. The described workflow comprises three steps. In the first step, an approach based on the design and analysis of computer experiments (DACE) is adopted in which data is generated for the parametrisation of Kriging models based on Latin hypercube sampling. In the second step, multi-objective optimisation is applied to the Kriging models. This study introduces an additional cost criterion, which includes the Rayleigh time step, in order to reduce the solution set size to one element. In the third step, the optimisation procedure is repeated with the actual DEM models, using the optimal parameter set from the Kriging models as the start value. This final step with the full DEM models refines the parameter set against experimental data. Since DEM material model calibration is time-consuming, the workflow is implemented into an automated process chain. In this paper, the methodology is described in detail and results are shown which illustrate the usefulness and effectiveness of this approach. Initial verification simulations run using the calibrated parameters give good agreement with experimental results. M. Rackl, C.D. Görnig, K.J. Hanley and W.A. Günthner

Challenges of Simulating Undrained Tests Using the Constant Volume Method in DEM

Liquefaction during earthquakes can cause significant infrastructural damage and loss of life, motivating a fundamental study of undrained sand response using discrete element modeling (DEM). Two methods are widely used in DEM for simulating the undrained response of soil. One approach is to numerically couple the DEM code with a fluid model. Alternatively, if the soil is fully saturated and water is assumed to be incompressible, the volume of the sample can be held constant to simulate an undrained test. The latter has the advantage of being computationally straightforward, but the assumption of a constant volume can cause some issues which are discussed in this paper. Depending on the contact model selected, extremely high deviatoric stresses and pore water pressures can be generated for dense samples using the constant volume approach which are not observed in corresponding laboratory tests. Furthermore the results of these constant volume simulations tend to be sensitive to the strain rate selected. T...

Verification of an Automated Work Flow for Discrete Element Material Parameter Calibration

Identification of parameter values for discrete element method (DEM) material models is a major issue for realistic simulation of bulk materials. Choosing suitable parameter values is often done using trial and error in a disorganized manner, where efficiency largely depends on the experience of the DEM user. A methodical work flow, which is based on Latin hypercube sampling, Kriging and numerical optimization, was composed with open-source software. The calibrated DEM materials were subsequently validated against the physical data from measurements and the number of required DEM simulations was recorded to assess the effectiveness of the overall method. The simulation results were within a few percent of the desired experimental values after an average of 14 DEM runs. Disadvantageous boundary conditions, like a wide factor value range or the optimum being located at an edge, did not considerably influence the quality of the results.

Use of elastic stability analysis to explain the stress-dependent nature of soil strength

The peak and critical state strengths of sands are linearly related to the stress level, just as the frictional resistance to sliding along an interface is related to the normal force. The analogy with frictional sliding has led to the use of a ‘friction angle’ to describe the relationship between strength and stress for soils. The term ‘friction angle’ implies that the underlying mechanism is frictional resistance at the particle contacts. However, experiments and discrete element simulations indicate that the material friction angle is not simply related to the friction angle at the particle contacts. Experiments and particle-scale simulations of model sands have also revealed the presence of strong force chains, aligned with the major principal stress. Buckling of these strong force chains has been proposed as an alternative to the frictional-sliding failure mechanism. Here, using an idealized abstraction of a strong force chain, the resistance is shown to be linearly proportional to the magnitude of the lateral forces supporting the force chain. Considering a triaxial stress state, and drawing an analogy between the lateral forces and the confining pressure in a triaxial test, a linear relationship between stress level and strength is seen to emerge from the failure-by-buckling hypothesis.

Correlation of Volume Ratio and Normalized Permittivity in Particle Mixture

The volume ratio in the total solid particle mixture is an important parameter indicating the characteristics of particulate mixtures. Therefore, a simple measurement method is desirable. In this paper, the normalized permittivities of two groups of particle mixtures are tested using electrical capacitance tomography (ECT) with the series and parallel models. The experimental results show that normalized permittivity changes obtained from the series ECT model have a very strong linear relationship with the volume ratio of higher permittivity materials in the mixture. The Maxwell Garnett formula can predict the normalized permittivity only after the permittivity of each type of particle is known. The comparison between two methods suggests that ECT with the series model is a better way to monitor the volume ratio in mixtures comprising two different types of solid particle.

Improving Estimates of Critical Time-Steps for Discrete Element Simulations

Discrete element modelling (DEM) is a powerful software tool for simulating the interactions of granular materials. An explicit, conditionally-stable time-stepping algorithm is typically used for the discrete element method which requires a time-step, either user-specified or computed internally within the code. It is well known that too small a time-step can slow down the simulations whereas too large a time-step can lead to numerical instability . In this work we use a model example to illustrate an alternative, more exact framework for calculating the critical stable time-step for an undamped collision. This framework can be readily extended to more complex configurations.

Closure to “Fabric and Effective Stress Distribution in Internally Unstable Soils” by T. Shire, C. O’Sullivan, K. J. Hanley, and R. J. Fannin

The authors have performed numerical tests to determine the distribution of fabric and effective stresses in internally unstable soils. Different grain size distribution curves as well as different densities were investigated. The authors have used the methods of Kezdi (1979) and Kenney and Lau (1985, 1986) to assess the internal stability. The relative density has a significant effect on the internal stability (Ahlinhan et al. 2012; ICOLD 2013). Accordingly, it is important to use a method takes into consideration the effect of density to assess the internal stability. Neither Kezdi’s method nor Kenney and Lau’s method takes the density in consideration. A suitable method is the one suggested by Dallo et al. (2013), which takes the effect of soil density in consideration. For instance, the soil (Gap med 18) is classified as border-line according to Kezdi’s method, whereas it is classified as unstable according to Kenney and Lau’s method, regardless of its relative density. According to Dallo et al. (2013), soil (Gap med 18) is unstable, transition (or border-line), and stable for the loose, medium, and dense relative densities, respectively. Also the internal stability prediction accuracy of Dallo et al. (2013)’s method is better than those of the Kezdi or Kenney and Lau methods (Dallo et al. 2013). The discusser used the method of Dallo et al. (2013) to assess the internal stability of the gap-graded soils tested by the authors (Table 1). It can be seen that the soils (Gap wide XX) are unstable, the soils (Gap narrow XX) are stable, whereas the internal stability of the soils (Gap med XX) depends on the relative density of the soil. Another discussion point is related to the authors’ adoption of the findings of Skempton and Brogan (1994) that the critical finer fraction (S ) falls between narrow limits of finer fractions by mass, Ffine 1⁄4 24–29% for dense and loose samples respectively. Also, the finer fraction (Smax) at which the finer particles completely separate the coarse particles from one another is given as Ffine 1⁄4 35%. The discusser believes that the critical finer fraction can be computed more accurately according to Indraratna et al. (2011), Eq. (1), or Dallo and Wang (2012), Eq. (2), as Smax 1⁄4 1 − nl 1 − nc nc ð1Þ

Incorporating the effect of pore pressure in undrained DEM simulations

Liquefaction in undrained soils coincides with the development of significant excess pore water pressures. The undrained behaviour of soils has been studied extensively using laboratory testing, but these tests cannot give any insight into the micromechanical changes that cause the observed macro-scale response. One way to obtain insight into the micromechanical behaviour is to use a suitable numerical technique such as the discrete element method (DEM). The ‘constant volume’ method, in which the sample volume is maintained constant throughout shearing, is often used to simulate the undrained test condition. In this method the soil sample is assumed to be perfectly saturated with an incompressible liquid. The constant volume method has the advantage of computational simplicity. However, some problems arise when simulating dense samples such as the generation of unrealistically high stresses and excessively large interparticle overlaps. There is a clear need to develop an alternative to constant volume simulations which retains the method’s computational efficiency but without the unphysicality. In this paper, several reasons are proposed for the inability of constant volume simulations to quantitatively capture a real soil’s undrained behaviour. Alternatives to the constant volume method are discussed, all of which allow the sample volume to vary during the simulation by incorporating the effect of pore pressure. One method was selected and implemented in the open-source LAMMPS DEM code, and its appropriateness for simulating undrained soil behaviour is explored with reference to monotonic simulations of sand.

Correlation analysis of solid particles' permittivity and composition using electrical capacitance tomography and Maxwell Garnett formula

The solid volume fraction and volume composition ratio of different solid particles in a silo are important parameters indicating the character of granular mixtures. In this paper, measurement of the two parameters is studied by utilizing Electrical Capacitance Tomography (ECT). This preliminary work focuses on comparing normalized permittivity changes from ECT with those predicted by the Maxwell Garnett formula (MGF) relating to solid volume fraction and volume composition ratio. Experimental results show that normalized permittivity changes from the two methods have a strong linear correlation in gas-solid two- and three-phase systems. This work demonstrates that ECT is a promising technique for monitoring solid volume fraction of single solid particles and the volume composition ratio of mixtures of two different solid particles in a silo.