INTERNATIONAL SOCIETY FOR SOIL MECHANICS AND GEOTECHNICAL ENGINEERING This paper was downloaded from the Online Library of the International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE). The library is available here: https://www.issmge.org/publications/online-library This is an open-access database that archives thousands of papers published under the Auspices of the ISSMGE and maintained by the Innovation and Development Committee of ISSMGE. 6/10 Predicting properties of young volcanic soils La prévision des propriétés des sols volcaniques jeunes F.G.THRALL, Project Engineer, Cornforth Consultants Inc., Portland, Oregon, USA J.R.BELL, Professor, Oregon State University, Corvallis, Oregon, USA SYNOPSIS: Engineering problems caused by fine-grained, poorly crystalline young volcanic soils have puzzled engineers for many years. Correlations developed from soil classification systems and labo­ ratory testing procedures for the more ordered crystalline soils do not apply to poorly crystalline soils. This paper reports a literature and laboratory investigation which identifies and develops correlations between selected index properties of young volcanic soils and corresponding strength, compaction, and consolidation parameters. The results are summarized by indicating correlation equations, coefficients of correlation, and comparisons with typical crystalline soil behavior. The results indicate that values for effective angle of internal friction and compression index can be estimated from void ratio. Undrained shear strength can be estimated using natural moisture content. Plastic limit provides an excellent index for estimating optimum water content. INTRODUCTION Geologically young fine-grained soils derived from volcanic material commonly contain confusing and ill-defined mixtures of crystalline clay min­ erals, poorly crystalline clay minerals, and rock- forming minerals. Such complex mixtures may in­ clude glass, halloysite, kaolinite, bundles of imogolite tubes, conglomerates of spherical allo- phane, and masses of gel-like amorphous materials. Figure 1 shows electron micrographs of typical allophane and imoglite assemblages. These soils have unusually high natural water contents, high liquid limits and plastic limits, and low plastic indices. They show different moisture density relationships before and after (a) (b) drying. Optimum water contents for these soils are usually below the natural water content Figure 1. Electron Micrographs of (a) Allophane (Newill, 1961). Particle size determinations from Alaska and (b) Imogolite from Japan. Scale are unreliable, depending upon the degree of dry­ is the same for both photographs. ing, and the treatment of the soil before and during testing (Birrell, 1966). Compressibility test results tend to underestimate field compres­ Construction Problems sion (Matyas, 1969). This paper reports a literature and laboratory Construction problems with andisol soils have investigation which summarizes correlations be­ puzzled geotechnical engineers for many years tween selected indices and engineering properties (Thrall, 1981). Soil classification systems and of young volcanic soils. correlations developed to predict the behavior of crystalline soils do not generally apply to young volcanic soils. Significant problems occur during Occurrence construction when these geotechnical behaviors are not recognized. Young volcanic soils occur abundantly in many In the natural state andisols are relatively parts of the world including Japan, Indonesia, stable, but when disturbed, tend to become a semi­ New Zealand, the Philippines and other Pacific liquid. Some may harden again after a short time. islands, Africa, Central and South America, They show an irreversible decrease in plasticity Hawaii, and the northwestern United States includ­ and volume upon drying. The compaction behavior ing Alaska. In the United States the term 'andi- is erratic and depends on the initial water con­ sol' is most commonly applied. In Japan they are tent and the amount of remolding induced during called 'humic allophane soils' or 'andisols'. In compaction. These behaviors are difficult to pre­ New Zealand they are 'yellow-brown loams', dict using standard soil descriptions and test 'yellow-brown pumice soils', or 'alvic' soils. procedures which typically require drying the In Chile and Argentina, they are 'trumao' or soils before testing. 'allophanic'. The term andisol is used in this paper. 555 6/10 <r> ûÿ .£ 50 & : 40 o IX A ° 20 a> D) K A c < Correlation for Crystalline Soils \ © 10 with Low Plasticity High Group (r=0 35) A r LU 0 0 1 2 3 4 5 6 7 8 Insitu Void Ratio te) Figure 2. Optimum Water Contents vs. Plastic Figure 3. Effective Angle of Internal Friction Limit vs. Insitu Void Ratio GEOTECHNICAL PROPERTIES Atterberg Limits Nine sites in the Oregon Cascade Range (Taskey, The Atterberg limits data plot well below the 1970) and two sites in Alaska (USA), located in A-line on the Casagrande Plasticity Chart. Liquid areas of geologically young volcanic ashfalls, limits range from 23 to 350 percent for the natu­ were sampled. Disturbed and thin wall tube sam­ ral condition, and 40 to 111 percent for previ­ ples of the upper 2 to 4 feet of soil were ob­ ously dried samples. Plasticity is greatly re­ tained at each site. Laboratory testing for duced upon drying. Plasticity loss is a continuum selected index properties and engineering param­ along which the amount of plasticity lost depends eters were carried out on these samples. upon the degree of drying. Soils with relatively Several hundred references containing geotechni- low natural water contents plot closer to the cal laboratory data on andisols were reviewed. A-line than those with higher natural water con­ Andisols were identified based on identification tents . by name or by physical-engineering properties, in Using methods described in Lumb (1966), paired, combination with location and climate. Atterberg undried liquid limit and plastic index data were limits and natural water content data were used as transformed to their standardized normal variates. the main indicators of andisols. Chi-square tests on the transformed data show that Approximately 200 soils worldwide were identi­ three normal distributions are evident. The mean fied from the literature and added to the test values for each of the three normal distributions results for the Oregon and Alaska soils. Complete are shown on Table I along with the corresponding results are tabulated in Thrall (1901). natural water content data. TABLE I Natural Water Content Mean Values for Liquid Limit, Plastic Index and Water Content Observations, by Group Reported natural water contents range from a low of 8 percent to a high of 313 percent with the Index Property Low Intermediate High majority from about 20 to 120 percent. Liquid Limit, % 64 116 180 Statistical tests of variance on the water con­ Plastic Index, % 15 38 64 tent data indicate that the natural water contents Water Content, % 38 111 >200 fall into two main groups: i) a low group with a mean water content of about 38 percent and stan­ dard deviation of about 12 percent; and ii) a high CORRELATIONS group with a mean of about 111 percent and stan­ dard deviation of about 35 percent. A third group Correlations between index properties and corre­ is indicated on the extreme high end of the water sponding strength, compaction, and consolidation content range; however, this represents a small parameters are shown on Figures 2, 3, 4 and 5. number of samples. These analyses assume the The correlations were obtained by developing natural water content distributions are normal. mathematical relationships between selected index Other investigators have shown this to be true properties and engineering parameters. Separate (Warkentin and Maeda, 1981). mathematical equations were developed for the Soils of the low water content group typically previously described low and high water content have red colors, and high silt-sand and halloy- groups. Soils with natural water contents less site percentage contents. yellow-brown colors, than 60 percent constitute the low group; the re­ high clay percentages, and high allophane percent­ mainder fall into the high group. The equations ages are typical of the high water content groups. were determined by regression analyses using the least squares method. The coefficients of corre­ lation are shown on the figures. Similar corre­ lations presented in the literature for crystal­ 556 6/10 Figure 4. Undrained Shear Strength vs. Natural Figure 5. Compression Index vs. Initial Void Water Content Ratio line soils are also reproduced on the figures. An increase of undrained shear strength with Table II summarizes the correlations for the andi- decreasing natural water content is shown on sols and the crystalline soils. Figure 4. For crystalline soils, this relation­ ship is unique for different clay minerals. The low water content andisol group agrees well with Moisture-Density the relationship for kaolinite (U.S. Navy, 1971) Dry density increases, and optimum water content decreases irreversibly with drying. This behavior Compressibility occurs continuously as the soil dries, much the same as with the Atterberg limits. Figure 5 shows a good logarithmic relationship Figure 2 shows a plot of optimum water content between compression index and void ratio. The versus plastic limit for soils which have not been regression curve follows the trend of the reported previously dried. The figure shows that a strong