Automating particle size and shape measurement in soil mechanics

Lead Research Organisation: Imperial College London
Department Name: Civil & Environmental Engineering

Abstract

It is well established that the geometry (size and shape) of the individual particles influences the overall response of a granular material to applied loads and deformations. Traditionally in soil mechanics we measured the size of soil particles by passing the particles through a succession of sieves with different aperture sizes. Shape was measured qualitatively, by comparing the shape of a small number of particles with standard (2D) charts. These methods have some disadvantages. Considering sieving, we have no information about the distribution of particle sizes between the standard sieve intervals. The results of a sieve analysis are not always consistent, especially when long platy particles are involved. Changes in particle size distribution are used to assess damage to the individual particles, and the small changes we need to measure cannot be reliably detected using a sieve analysis. Regarding shape analysis, we know that particles in granular materials are three-dimensional. We also know that to get a statistically valid assessment of particle shape and particle shape variation in a granular material, the shape of thousands of particles should be considered. The traditional simple, visual description of particle geometry does not allow objective, quantitative, three-dimensional shape analysis of statistically meaningful numbers of particles. A new technology, the QicPic apparatus manufactured by Sympatec, includes a system for dispersing particles, a high speed camera and sophisticated image analysis software to measure particles sizes and shapes. This apparatus overcomes the limitations of the traditional low tech approaches, allowing 3D assessment of size and shape for large numbers of particles. We have had access to this apparatus, it was installed in our laboratory for a one-week trial period. During this period we established that the apparatus can produce valid measurements for the particles we are interested in, it can easily be used by our research students, and the degree of sophistication of the output (including black and white images of individual particles) makes it a good tool for fundamental research on particle morphology.This apparatus would provide us with improved capabilities for quality control for a range of experimental work involving granular materials in the department. These materials include natural and artificial sands used in soil mechanics research, morsellised bone used in biomedical engineering research and artificial aggregates that are being developed from waste by-products as environmental engineering research. The apparatus can also be used to better understand how the individual particles are damaged when a granular material is loaded and deformed, and this understanding is important for developing numerical models of the material. The apparatus would also be useful to support ongoing research that will develop a relationship between the overall material response and the characteristics of the constituent particles.The applicants are already collaborating to develop methods to quantify both particle strength and particle geometry and relate these characteristics to the overall material response. Our current facilities include an optical microscope,a surface mapping microscope, and single particle mechanical apparatii. This apparatus would complement these tools, and strengthen our expertise in this area.
 
Description Overall we have been using this apparatus in the soil mechanics laboratory at Imperial College since it arrived in 2008. The apparatus has been extensively used in undergraduate (MEng) and postgraduate (MSc, PhD) research. It represents a significant improvement over sieving as small amounts of material can be considered and a large amount of detailed information on the particle morphologies become available.



On particle shape in general:

1. The shape measures of sphericity and convexity that are output via QicPic are less subjective than the previous approach of describing particles as rounded, subrounded, subangular and angular. However there is a clear correlation between the conventional qualitative measures and the quantitative QiPic measures. There is therefore a lot of scope for this method to be integrated in both the research and practice sectors of geotechnical engineering.

2. The QicPic data consider a number of size measures and none of these is directly applicable to conventional sieving, however the Feret min measure of size seems to best approximate sieve data.

3. We compared the QicPic size measures with 3D morphological analysis of Reigate sand (Fonseca et al., Soils and Foundations) and found differences as a consequence of 2D nature of information.



On particle breakage:



1) The better accuracy of the QicPic compared to sieves highlighted for the first time how in compression breakage is not constant at a given stress level for any initial density as had previously been thought. This makes our models of soil behaviour rather more complex as the current grading will now be a function of stress and initial density.

2) Data published by Coop and Altuhafi obtained using the QicPic highlight for the evolution from a "standard" sand mode of behaviour in compression for poorly graded sands (with well-defined yield at the onset of breakage and unique normal compression line) to a transitional mode for very well graded sands (with no unique normal compression line). The transitional mode was found not to give any measurable particle breakage in sands, despite the large volumetric strains in compression. Some aspects of this had been seen in earlier work, but this was the first time that the complete evolution with grading had been seen. The work also confirmed this evolution was the same for three sands of very different mineralogies, including a quartz sand (i.e. the most common type).

3) The shape analysis on the QicPic allowed the nature of changes to particle morphology as the particles broke to be highlighted.



Soil structure interaction:

The QicPic enabled detailed analysis of grain crushing around piles and showed that large scale crushing takes place and that this dominates the interface shearing response during pile installation. The crushed material is concentrated within "shear zones". These are about 4 - 13 times the median particle diameter.

The QicPic enables a study of fines migration during shearing.
Exploitation Route The research will have a long term impact on geotechnical engineering practice via the development of improved constitutive models for soil response.



This is an effective and easy to use approach for quantifying sand morphology. It could easily be adopted within commercial geotechnical laboratories. This was an equipment grant and the QicPic apparatus purchased in this grant will continue to be used to under pin research at Imperial College.



The findings listed above will hopefully

a) lead to a move away from subjective, qualitative assessment of sand particle shape in geotechnical research and practice

b) inform development of constitutive models that consider particle breakage

c) lead to better understanding of interface friction for piles
Sectors Construction