000004498 001__ 4498
000004498 005__ 20141118192700.0
000004498 0177_ $$2doi$$a10.3850/978-981-07-2219-7_P159

000004498 0247_ $$210.3850/978-981-07-2219-7_P159
$$adoi
000004498 04107 $$aeng
000004498 046__ $$k2012-05-23
000004498 100__ $$aChing, Jianye
000004498 24500 $$aOverall Shear Strength of Soil Mass With Spatial Variability

000004498 24630 $$n5.$$pProceedings of the 5th Asian-Pacific Symposium on Structural Reliability and its Applications
000004498 260__ $$bResearch Publishing, No:83 Genting Lane, #08-01, Genting Building, 349568 SINGAPORE
000004498 506__ $$arestricted
000004498 520__ $$2eng$$aSpatial variability is a realistic phenomenon for soil shear strengths. Based on statistical analysis, Vanmarcke (1977) showed that the averaged property of a random field over a region has mean value identical to the inherent mean, while the variance of the average is less than the inherent variance. Vanmarcke's derivations for spatial averaging are purely statistical, not involving mechanical principles, e.g., force equilibrium, compatibility, constitutive law, etc. However, the "overall" shear strength of a soil mass should be governed by the mechanical principles. The purpose of this study is therefore to understand the mechanisms that govern the overall shear strength. Random field finite element analyses (FEA) are conducted. The spatially varying shear strength τ<sub>f</sub> is simulated by stationary Gaussian random fields. When assigning the simulated τ<sub>f</sub> to each element, the local averaging subdivision algorithm developed by Fenton and Vanmarcke (1990) is taken to conduct element-level averaging within each element. What follow are the important findings:
  1. Although the spatial averaging of shear strength over a domain D (τ<sub>f</sub><sup>D</sup>) has similar statistical behaviors to those predicted by Vanmarcke’s theory, the statistical behaviors of the overall shear strength (τ<sub>f</sub><sup>O</sup>) are NOT consistent to those predicted by Vanmarcke’s theory. In fact, the mean value of τ<sub>f</sub><sup>O</sup> is always less than that of τ<sub>f</sub><sup>D</sup>, and the coefficient of variation (COV) of τ<sub>f</sub><sup>O</sup> is always larger than that of τ<sub>f</sub><sup>D</sup>. 
 2. τ<sub>f</sub><sup>O</sup> is very close to the line average along the actual slip curve (τ<sub>f</sub><sup>S</sup>). The mechanism for τ<sub>f</sub><sup>O</sup> is nearly the same as that for τ<sub>f</sub><sup>S</sup> – τ<sub>f</sub><sup>O</sup> is the result of taking line averaging along the potential slip curves and then taking the minimum value among these averages, rather than taking spatial average of the entire domain D. Due to the action of taking minimum, τ<sub>f</sub><sup>S</sup> (or τ<sub>f</sub><sup>O</sup>) may have mean value less than that for τ<sub>f</sub><sup>D</sup>, especially when the line averaging effect along the potential slip curve is weak, and may have COV larger than that for τ<sub>f</sub><sup>D</sup>.

000004498 540__ $$aText je chráněný podle autorského zákona č. 121/2000 Sb.
000004498 653__ $$a

000004498 7112_ $$a5th Asian-Pacific Symposium on Structural Reliability and its Applications$$cSingapore (SG)$$d2012-05-23 / 2012-05-25$$gAPSSRA2012
000004498 720__ $$aChing, Jianye$$iPhoon, Kok-Kwang
000004498 8560_ $$ffischerc@itam.cas.cz
000004498 8564_ $$s236515$$uhttps://invenio.itam.cas.cz/record/4498/files/P159.pdf$$yOriginal version of the author's contribution as presented on CD, .
000004498 962__ $$r4180
000004498 980__ $$aPAPER