THESIS
2014
iv leaves, v-xxii, 226 pages : illustrations, map ; 30 cm
Abstract
Vegetation is well recognized to provide mechanical reinforcement to soils, which increases soil apparent shear strength and hence slope stability. The root-water uptake process, the soil suction distribution in vegetated soils and their impacts on the stability of a vegetated slope during rainfall, however, are not well understood. Vegetation is also used to prevent surface erosion in an environmentally friendly manner. In particular, the effectiveness of Hong Kong native grass species in controlling the erodibility of completely decomposed granite (CDG) compacted at different densities has not been quantitatively evaluated. The native grass species that control erosion effectively for Hong Kong soil slopes should also be selected in a scientific manner.
Natural or engineered soil mas...[
Read more ]
Vegetation is well recognized to provide mechanical reinforcement to soils, which increases soil apparent shear strength and hence slope stability. The root-water uptake process, the soil suction distribution in vegetated soils and their impacts on the stability of a vegetated slope during rainfall, however, are not well understood. Vegetation is also used to prevent surface erosion in an environmentally friendly manner. In particular, the effectiveness of Hong Kong native grass species in controlling the erodibility of completely decomposed granite (CDG) compacted at different densities has not been quantitatively evaluated. The native grass species that control erosion effectively for Hong Kong soil slopes should also be selected in a scientific manner.
Natural or engineered soil masses exhibit complex spatial heterogeneity patterns, which have not been sufficiently captured and modelled so far. Particularly the spatial variability of plant characteristics has not been considered in previous research. It is still rather challenging to apply existing random field theory to characterise root-water uptake and simulate stability of vegetated slopes.
The principal objectives of this research are to (1) extend random field theory to characterise geotechnical anisotropic spatial variation, simulate the spatial variation of permeability function and plant transpiration, and apply the theory to green slopes; (2) propose a general methodology for characterising multivariate geotechnical random fields; (3) propose a method for quantifying root characteristics (i.e., root depth, volume and mass) and investigating the effect of soil compaction on the root characteristics of three Hong Kong native grass species; (4) design a field experimental programme for verifying the numerical scheme and quantitatively assessing the erosion resistance of soils vegetated with local grass, through which an optimum grass species that controls erosion effectively is recommended for Hong Kong soil covers.
Random field theory that characterises the spatial correlation of a target physical property is first extended to fully account for actual natural scenarios where the major and minor directions of spatial variation may not be perpendicular. An algorithm is developed to extend the existing random field theory to simulate this kind of spatial variation. Five types of random fields of a soil property including isotropy, transverse anisotropy, rotated anisotropy, general anisotropy and general rotated anisotropy are generated based on the extended theory. They can be further applied as inputs for seepage or stress-strain analysis. Theoretical formulations of scale of fluctuation as a function of directional angle are derived for the five basic patterns of anisotropy. Applications of the extended random field theory into bare and vegetated slopes are illustrated. Both the permeability function for the soil and the maximum transpiration rate of Cynodon Dactylon (Bermuda Grass) are considered as random fields. The conditions of soils, rainfall and vegetation under which matric suction can be retained and how vegetation helps stabilise the slope when subject to rainfall of a certain period are demonstrated through the proposed numerical scheme.
A general methodology is proposed for generating multivariate cross-correlated random fields. The joint distribution is rigorously established with the aid of a copula function that describes the dependence structure among the individual variables. Then independent random fields are generated through Cholesky decomposition. Afterwards, cross-correlated random fields are established by modifying the independent random fields using the copula function.
A comprehensive field-testing programme was designed and implemented to verify the numerical scheme regarding the suction changes and to test the erosion resistance of grassed soils. A method is proposed for quantitatively assessing the erodibility of a grass-covered CDG soil bed, which is commonly found in Hong Kong. The soil bed and plantation scheme were designed by the author and her classmates. Three types of Hong Kong native turf grass including Cynodon Dactylon, Paspalum Notatum and Zoysia Matrella were planted on three soil grounds with degrees of compaction of 80%, 90% and 100%, respectively. Comparison of suctions retained after rainfall in the vegetated soil beds between numerical solutions and field measurements shows good agreement. The featural parameters of grass roots on each compacted ground including root mass density, root volume density and root depth were measured in two growth stages. A jet index apparatus was modified and applied to measure the erodibility properties (i.e., coefficient of erodibility and critical shear stress) of these vegetated soils in the two test stages. Results show that Cynodon Dactylon and Zoysia Matrella have higher root mass density values than Paspalum Notatum does and reduce the susceptibility of soil erosion more effectively, hence are recommended for use in green slope engineering in Hong Kong.
Post a Comment