13 Foundations on Difficult Soils Introduction In many areas of the United States and other parts of the world, certain soils make the construction of foundations extremely difficult. For example, expansive or collapsible soils may cause high differential movements in structures through excessive heave or settlement. Similar problems can also arise when foundations are constructed over sanitary landfills. Foundation engineers must be able to identify difficult soils when they are encountered in the field. Although not all the problems caused by all soils can be solved, preventive measures can be taken to reduce the possibility of damage to structures built on them. This chapter outlines the fundamental properties of three major soil conditions—collapsible soils, expansive soils, and sanitary landfills—and methods of careful construction of foundations. Collapsible Soil Definition and Types of Collapsible Soil Collapsible soiis, which are sometimes referred tv as metastable soils, arc unsaturated soils that undergo a large change in volume upon saturation. The change may oF May not be the result of the application of additional load. The behavior of collapsing soils under load is best explained by the typical void ratio effective pressure plot (¢ against Jog 0’) for a collapsing soil, as shown in Figure 13.1. Branch ab is determined from the consolidation test on a specimen at its natural moisture content. At an effective Pre sure level of a, the equilibrium void ratio is e,. However, if water is introduc int, the specimen for saturation, the soil structure will collapse. After saturation, the ea librium void ratio at the same effective pressure level or, is e2: cd is the branch of e-log o’ curve under additional load after saturation, Foundations that are consttuc on such soils may undergo large and suclden settlement if the soil under them become saturated with an unanticipated supply of moisture. The moisture may come fo" ee of several sources, such as (a) broken water pipelines, (b) leaky sewers, () drainags13.3 Effective pressure, 0” (log scale) Figure 13.1 Nature of variation of void ratio with pressure for a collapsing soil from reservoirs and swimming pools, (d) a slow increase in groundwater, and so on, This type of settlement generally causes considerable structural damage. Hence, iden- fying collapsing soils during field exploration is crucial The majority of naturally occurring collapsing soils are aeolian—that is, wind- deposited sands or silts, such as loess, aeolic beaches, and volcanic dust deposits. The deposits have high void ratios and low unit weights and are cohesionless or only slightly cohesive. Loess deposits have silt-sized particles. The cohesion in loess may be the result of clay coatings surrounding the silt-size particles. The coatings hold the particles in a rather stable condition in an unsaturated state. The cohesion also may be caused by the presence of chemical precipitates leached by rainwater. When the soil becomes saturated, the clay binders lose their strength and undergo a structural collapse. In the United States, large parts of the Midwest and the arid West have such types of deposit. Loess deposits are also found over 15 to 20% of Europe and over large parts of China. The thickness of loess deposits can range up to about 10 m (33 ft) in the central United States. In parts of China it can be up to 100 m (330 ft). Figure 13.2 shows the extent of the loess deposits in the Mississippi River basin, Many collapsing soils may be residual soils that are products of the weathering of parent rocks. Weathering produces soils with a large range of particle-size distribution. Soluble and colloidal materials are leached out by weathering, resulting in large void ratios and thus unstable structures. Many parts of South Africa and Zimbabwe have residual soils that are decomposed granites. Sometimes collapsing soil deposits may be left by flash floods and mudflows. These deposits dry out and are poorly consolidated. An excellent review of collapsing soils is that of Clemence and Finbar (1981). Physical Parameters for Identification Several investigators have proposed various methods for evaluating the physical parame ters of collapsing soils for identification, Some of these methods are discussed briefly in Table 13.1. Jennings and Knight (1975) suggested a procedure for describing the collapse potential of a soil: An undisturbed soil specimen is taken at its natural moisture con- tent in a consolidation ring. Step loads are applied to the specimen up to a pressure level or), of 200 N/m? (~29 Ib/in?). (In Figure 13.1, this is a7,.) At that pressure, theSouth Dakota Figure 13.2 Loess deposit in Mississippi River basin specimen is flooded for saturation and left for 24 hours. This test provides the void ratios e, and e; before and after flooding, respectively. The collapse potential may now be calculated as C, «igs «Bay le, where jatural void ratio of the soil vertical strain & Ae ‘The severity of foundation problems associated with a collapsible soil have been correlated with the collapse potential C,, by Jennings and Knight (1975). They were summarized b¥ ‘Clemence and Finbarr (1981) and are given in Table 13.2. Holtz and Hilf (1961) suggested that a loessil soil that has a void ratio large enoueh '® allow its moisture content to exceed its liquid limit upon saturation is susceptible to cola So, for collapse, W(aweareay > LL t oowhere LL = liquid limit. However, for saturated soils, €, = wG, (13.3) where G, = specific gravity of soil solids. Table 13.1 Reported Criteria for Identification of Collapsing Soil* Criteria Coet ient of subsidence: void ratio at liquid limit fe natural void ratio K = 05-0.75: highly collapsible K = 1.0: noncollapsible loam K = 15-20: noncollapsible soils Clevenger 198 dry unit weight is less than 12.6 KN/m'(=80 Ih/fi*). settlement will be large; if dry unit weight is greater than 14 N/m’ (=90 lb/ft) settlement will be small. | natural moisture content — plastic limi Priklonski 19820 Ky =§ ore: plasticity index Kp <0: highly collapsible soils Kp > 0.5; noncollapsible soils Ky > 1.0: swelling soits saturation moisture content liquid limit This was put into graph form, ee 1+e, where ¢, = natural void ratio and e;, = void ratio at liquid limit. For natural degree of saturation less than 60%, if L. > ~0.1, the soil is a collapsing soil. wy PL Feda 1964, = “2 BE Gibbs 1961 Collapse ratio, R Soviet Building Code 1962 natural water content, S, = natural degree of saturation, PL = plastic limit, and PI = plasticity index. For S, < 100%, if K,, > 0.85, the soil is a subsident soil. Benites 1968 A dispersion test in which 2 g of soil are dropped into 12 ml of distilled water and specimen is timed Until dispersed: dispersion times of 20 to 30 s were ‘obtained for collapsing Avizona soils Handy 1973 Towa loess with clay (<0.002 mm) contents: ‘<16%: high probability of collapse 16-24%: probability of collapse 24-32%: less than 50% probability of collapse >32%: usually safe from collapse “Modified after Lutenegger and Saber (1988)Table 13.2 Relation of Collapse Potential to the Severity of Foundation Problems! G, 1%) Severity of problem 0-1 No problem a 1s Moderate trouble 510 ‘Trouble 10-20 Severe trouble 20 Very severe trouble "From Clemence, S. P., and Finbarr, A. 0. (1981). “Design Considerations for Collapsible Sas Journal of the Geotechnical Engineering Division, American Society of Civil Engineers, Vol. 197, GT3 pp. 305-317. With permission from ASCE. i Combining Eqs. (13.2) and (13.3) for collapsing soils yields & = (LL)(G,) (13.4) The natural dry unit weight of the soil required for its collapse is Gow Ge w= THe, 14 (LL)\(G) (13.5) For an average value of G, = 2.65, the limiting values of yj for various liquid limits may now be calculated from Eq, (13.5). Figure 13.3 shows a plot of the preceding limiting values of dry unit weights against the corresponding liquid limits. For any soil, if the natural dry unit weight falls below the limiting line, the soil is likely to collapse. Loessial soil likely to collapse Natural dry unit weight, y, (kNm') 10 lo 20 30 40 45. Figure 13.3 Loessial soil likely Liquid limit to collapseCare should be taken to obtain undisturbed samples for determining the collapse potentials and dry unit weights—preferably block samples cut by hand. The reason is that samples obtained by thin-walled tubes may undergo some compression during the impling process. However, if cut block samples are used, the boreholes should be made without water. Procedure for Calculating Collapse Settlement Jennings and Knight (1975) proposed the following laboratory procedure for determining the collapse settlement of structures upon saturation of soil: Step 1, Obtain two undisturbed soil specimens for tests in a standard consolidation test apparatus (oedometer). Step 2. Place the two specimens under 1 kN/m?(0.15 Ib/in’) pressure for 24 hours. Step 3. After 24 hours, saturate one specimen by flooding. Keep the other speci- men at its natural moisture content, Step 4. After 24 hours of flooding, resume the consolidation test on both specimens by doubling the load (the same as in the standard consolidation test) to the desired pressure level. Step 5. Plot the e-log a" graphs for both specimens (Figures 13.4a and b) Step 6. Calculate the in situ effective pressure, «7, Draw a vertical line correspond- ing to the pressure a, Step 7, Prom the e-log c'}, curve of the soaked specimen, determine the precunsol- lation pressure, o. If //or, = 0.8-1.5, the soil is normally consolidated; however, if o//o7, > 1.5, the soil is preconsolidated. > log a" Slog a Specimen at natural ‘Specimen at cases natural content ; moisture content Void ratio, € Void ratio, e Secor @ ) Figure 13.4 Settlement calculation from double oedometer test: (a) normally consolidated soil; (b) overconsotidated soilStep 8. Determine e;, corresponding to a, from the elog or, curve of the specimen, (This procedure for normally consolidated and overconso} soils is shown in Figures 13.4a and b, respectively.) ‘tated Step 9% ‘Through point (c7,€;) draw a curve that is similar to the e-log a? g obtained from the specimen tested at its natural moisture content,” U® Step 10. Determine the incremental pressure, Ar’, on the soil caused by the cong tion of the foundation. Draw a vertical line corresponding to the of, + Ao’ inthe e-log a” curve. Step 11. Now, determine Ae; and Ae). The settlement of soil without change natural moisture content is 7 tru. Pressure of in the (13.6) where H = thickness of soil susceptible to collapse. Also, the settlement caused by collapse in the soil structure is Aes Sia = 73 @ A) (13.7) Foundation Design in Soils Not Susceptible to Wetting For actual foundation design purposes, some standard field load tests may also be con- ducted. Figure 13.5 shows the relation of the nature of load per unit arca versus settle- ment in a field load test in a loessial deposit. Note that the load-settlement relationship is essentially linear up to a certain critical pressure, 7, at which there is a breakdown of the soil structure and hence a large settlement. Sudden breakdown of soil structure is more common with soils having a high natural moisture content than with normally dry soils. If enough precautions are taken in the field to prevent moisture from increasing under structures, spread foundations and mat foundations may be built on collapsible Load per unit area Figure 13.5 Field load test in loessial soil: 88 pet unit area versus settlementfog ruwnuauor uesigt nt oune 1 Susvepumie tw rrewny Oa .. However, the foundations must be proportioned so that the critical stresses (see Figure 13.5) in the field are never exceeded. A factor of safety of about 2.5 to 3 should be used to calculate the allowable soil pressure, or Ge FS where 4) = allowable soil pressure FS = factor of safety ‘The differential and total settlements of these foundations should be similar to those of foundations designed for sandy soils, Continuous foundations may be safer than isolated foundations over collapsible soils in that they can effectively minimize differential settlement. Figure 13.6 shows a typical procedure for constructing continuous foundations. This procedure uses footing beams and longitudinal load-bearing beams. In the construction of heavy structures, such as grain elevators, over collapsible soils, settlements up to about 0.3 m (=I ft) are sometimes allowed (Peck, Hanson, and Thornburn, 1974). In this case, tilting of the foundation is not likely to 0 because there is no eccentric loading. The total expected settlement for such structures can be estimated from standard consolidation tests on specimens of field moisture content. Without eccentric loading, the foundations will exhibit uniform settlement over loessial deposits; however, if the soil is of residual or colluvial nature, settlement may not be uniform. The reason is the nonuniformity generally encountered in resid- ual soils. Extreme caution must be used in building heavy structures over collapsible soils. If large settlements are expected, drilled-shaft and pile foundations should be considered. ‘These types of foundation can transfer the load to a stronger load-bearing stratur Load bearing beams, Figure 13.6 Continuous foundation with load-bearing beams (From Clemence, S. P., and Finbarr, A. 0. (1981). “Design Considerations for Collapsible Soils,” Journal of the Geotechnical Engineering Division, American Society of Civil Engineers, Vol. 107, GT3 pp. 305-317. With permission from ASCE.)Foundation Design in Soils Susceptible to Wetting IF the upper layer of soil i likely to get wet and collapse sometime after construction the foundation, several design techniques to avoid failure of the foundation may ben sidered. They are as follow: oti Dynamic Compaction If the expected depth of wetting is about 1.5 t0 2 m (~5 to 6.5 ft) from the ground up face, the soil may be moistened and recompacted by heavy rollers. Spread footings ang ‘mats may be constructed over the compacted soil. An alternative to recompaction by hea rollers is heavy tamping, which is sometimes referred to as dynamic compaction. (See Chapter 14.) Heavy tamping consists primarily of dropping a heavy weight repeatedly on the ground. The height of the drop can vary from 8 to 30 m (~25 to 100 ft). The stress ‘waves generated by the dropping weight help in the densification of the soil Lutenegger (1986) reported the use of dynamic compaction to stabilize a thick layer of friable loess before construction of a foundation in Russe, Bulgaria. During field exploration, the water table was not encountered to a depth of 10 m (33 ft), and the natural moisture content was below the plastic limit. Initial density measurements made on undisturbed soil specimens indicated that the moisture content at saturation would exceed the liquid limit, a property usually encountered in collapsible loess. For dynamic compaction of the soil, the upper 1.7 m (5.6 ft) of crust material was excavated. A circular concrete weight of 133 kN (~15 ton) was used as a hammer. At each grid point, compaction was achieved by dropping the hammer 7 to 12 times through a vertical distance of 2.5 m (8.2 ft). Figure 13.7 shows the dry unit weight of the soil before and after compaction. The increase in dry unit weight of the soil shows that dynamic compaction can be used effec tively to stabilize collapsible soil. Dry unit weght (kN/m?) 4 16 Is 20 7 T T Before compaction After ‘compaction Depth (m) Figure 13.7 Dynamic compaction of # friable loess layer in Russe, Bulgaria a 1 L (Adapted after Lutenegger, 1986)Chemical Stabilization If conditions are favorable, foundation trenches can be flooded with solutions of sodium sili- cate and calcium chloride to stabilize the soil chemically. The soil will then behave like a soft sandstone and resist collapse upon saturation. This method is successful only if the solutions can penetrate to the desired depth; thus, it is most applicable to fine sand deposits. Silicates ‘are rather costly and are not generally used. However, in some parts of Denver, silicates have been used successfully “The injection of a sodium silicate solution to stabilize collapsible soil deposits has been used extensively in the former Soviet Union and Bulgaria (Houston and Houston, 1989). This process, which is used for dry collapsible soils and for wet collapsible soils that are likely to compress under the added weight of the structure to be built, consists of three steps: Step 1. Injection of carbon dioxide to remove any water that is present and for preliminary activation of the soil Step 2. Injection of sodium silicate grout Step 3. Injection of carbon dioxide to neutralize alkalies. Vibrofloation and Ponding When the soil layer is susceptible to wetting to a depth of about 10 m (~30 ft), several techniques may be used to cause collapse of the soil before construction of the foundation is begun. Two of these techniques are vibroflotation and ponding (also called flooding). Vibroflotation is used successfully in free-draining soil. (See Chapter 14.) Ponding—by constructing low dikes—is utilized at sites that have no impervious layers. However, even after saturation and collapse of the soil by ponding, some additional settlement of the soil may occur after construction of the foundation is begun. Additional settlement may also be caused by incomplete saturation of the soil at the time of construction. Ponding may be used successfully in the construction of earth dams, Extending Foundation Beyond Zone of Wetting If precollapsing the soil is not practical, foundations may be extended beyond the zone of, possible wetting, although the technique may require drilled shafts and piles. The design Of drilled shafts and piles must take into consideration the effect of negative skin friction resulting from the collapse of the soil structure and the associated settlement of the zone of subsequent wetting In some cases, a rock-column type of foundation (vibroreplacement) may be consid- ered. Rock columns are built with large boulders that penetrate the collapsible soil layer. They act as piles in transferring the load to a more stable soil layer. General Nature of Expansive Soils Many plastic clays swell considerably when water is added to them and then shrink with the loss of water. Foundations constructed on such clays are subjected to large uplifting forces caused by the swelling. These forces induce heaving, cracking, and the breakup of both building foundations and slab-on-grade members. Figure 13.8 showsV7 se cerca aa Figure 13.8 Cracks in a wall due to heaving of an expansive clay (Courtesy of Anand Puppata, University of Texas at Arlington, Arlington, Texas) the cracks in a wall due to excessive heaving. Expansive clays cover large parts of the United States, South America, Africa, Australia, and India. In the United States, these clays are predominant in Texas, Oklahoma, and the upper Missouri Valley. In general, expansive clays have liquid limits and plasticity indices greater than about 40 and 15, respectivel¥, AAs noted, the increase and decrease in moisture content causes clay to swell and shrink. Figure 13.9 shows shrinkage cracks on the ground surface of a clay weathered fom the Eagle Ford shale formation in the Dallas-Fort Worth, Texas area. The depth in a soil © which periodic changes of moisture occur is usually referred to as the active cone (S°E Figure 13.10). The depth of the active zone varies, depending on the location ©! : Some typical active-zone depths in American cities are given in Table 13.3. In some clas and clay shales in the westem United States, the depth of the active zone can be a8 ™E 0 ft). The active-zone depth can easily be determined by plotting the . ity index against the depth of the soil profile over several seasons. Figure 13.1 shows *® a plot for the Beaumont formation in the Houston area. the sitesro werrerar ature OF Expansive Sous Figure 13.9 Shrinkage cracks on ground surface in a clay weathered from Eagle Ford shale formation in the Dallas-Fort Worth area (Courtesy of Thomas M. Petry, Missouri University of Science und Technology, Rolla, Missouri) Grunt sn 4 4 3 | ben Figure 13.10 Definition of active zone 697Table 13.3 Typical Active-Zone Depths in Some US. Cities th of active zor City (fo) Houston 15103 51010 Dallas 211046 71015 San Antonio 3109 10030 Denver 31046 10015 “Alter O'Neill nd Poormoayed (1980) (O'Neill, M. W.,, and Poormoayed, N.(1980). “Methodology for Foundations on Expansive Clays,” Journal of the Geotechnical Engineering Division, American Society of Civil Engineers, Vol. 106, No. GT12, pp. 1345-1367. With permission from ASCE.) Shrinkage cracks can extend deep into the active zone, Figure 13.12 shows intercon. nected shrinkage cracks extending from the ground surface into the active zone in an expansive clay. ‘To study the magnitude of possible swell in a clay, simple laboratory oedometer tests can be conducted on undisturbed specimens. Two common tests are the unrestrained swell test and the swelling pressure test. They are described in the following sections. Liquidity index -1 0 + 1 “ Approximate, : depth / of seasonal —} > |_ change. 167m | ! i i 4 lent & | q zs 1 i a | ' Range over Ft sever ' | seasons i ouston 4 ! Figura 13.11 Active zone in Hou! H i ‘area, Beaumont formation (O'Neill, M- esas and Poormoayed, N. (1980). t 1 “Methodology for Foundations 0" 5 i Expansive Clays,” Journal of the ' t Geotechnical Engineering Division i American Society of Civil Ensineet® I Vol. 106, No. GT12, pp. 1345-1367 6 . With permission from ASCE.)Interconnected shrinkage cracks extended from the ground surface into the active zone (Courtesy of Thomas M. Petry, Missouri University of Science and Technology Rolla, Missouri) In the unrestrained swell test, the specimen is placed in an oedometer under a small surcharge of about 6.9 kN/m: Water is then added to the specimen, and the expansion of the volume of the specimen (i.., its height; the area of cross section is constant) is measured until equilibrium is reached. The percentage of free swell may than be expressed as the ratio AH Swifice)(%) = =~ (100) (13.9) where Su(ine) = free swell, as a percentage AH = heightof swell due to saturation H = original height of the specimen Vijayvergiya and Ghazzaly (1973) analyzed various soil test results obtained in this manner and prepared a correlation chart of the free swell, liquid limit, and natural moisture content, as shown in Figure 13.13. ONeill and Poormoayed (1980) developed a relationship for calculating the free surface swell from this chart AS = 0.003325 ,100) (13.10)20 10 Liquid limit 70 Percent swell, Seine) Figure 13.13 Relation between percentage of free swell, liquid limit, and natural moisture content (From Vijayvergiya, V. N. and Ghazzaly, 0.1, (1973), “Prediction of Swelling Potential of Natural Clays,” Proceedings, ‘Third International Research and Engineering Conference on Expansive Clays, pp. 227-234. With permission ou 0 0 =20 3050 Natural moisture content (%) from ASCE.) where AS, = free surface swell Z = depth of active zone Su(iee) = free swell, as a percentage Swelling Pressure Test ‘The swelling pressure can be determined from two different types of tests. They * Conventional consolidation test * Constant volume test The two methods are briefly described here. Conventional Consolidation Test In this type of test the specimen is placed in a oedometer under @ small surcharge of 6.9 kN/m (1 Ib/in’). Water is added to the specimen, allowing it to swell and reach an eauili- rium position after some time. Subsequently, loads are added in convenient steps. and the specimen is consolidated. The plot of specimen deformation (6) versus log a" is shown Mt Figure 13.14, The 8 versus log a” plot crosses the horizontal line through the point of init condition. The pressure corresponding tothe point of intersection isthe zero swell press: 7 aboutSpecimen, deformation, 5 Consolidation | to addition of ater “Initial condition Figure 13.14 Zero swell pressure from Pressure, @" conventional consolidation test Constant Volume Test ‘The constant volume test can be conducted by taking a specimen in a consolidation ring and applying a pressure equal to the effective overburden pressure, «7, plus the approximate anticipated surcharge caused by the foundation, cr. Water is then added to the specimen. the specimen starts to swell, pressure is applied in small increments to prevent swelling, Pressure is maintained until full swelling pressure is developed on the specimen, at which time the total pressure is Om = 0 of bof 3.11) where 4 = total pressure applied to prevent swelling, of zero swell pressure 0, = additional pressure applied to prevent swelling after addition of water Figure 13.15 shows the variation of the percentage of swell with effective pressure during a swelling pressure test. (For more information on this type of test, see Sridharan et al., 1986.) Unloading Swell, s, (9%) Figure 13.15 Swelling pressure Effective pressure testA oof about 20 to 30 KN/m? (400 to 650 Ib/ft*) is considered to be low, ang @ Of 1500 to 2000 kN/mn* (30,000 to 40,000 Ib/ ft) is considered to be high. After ym swell pressure is attained, the soil specimen can be unloaded in steps to the level of re effective overburden pressure, The unloading process will cause the specimen re swell. The equilibrium swell for each pressure level is also recorded. The Variation of swell, sin pereent, and the applied pressure on the specimen will be like that shown 4 Figure 13.15 ‘The constant volume test can be used to determine the surface heave, AS. for a fouy. dation (O’Neill and Poormoayed, 1980) as given by the formula AS = Disieqs) (%)](H,) (0.01) (13.12) a where Spi) = swell, in percent, for la AH, = thickness of layer i It is important to point out that the zero swell pressure (c7,,) obtained from the conventional consolidation test and the constant volume test may not be the same. Table 13.4 summarizes some laboratory test results of Sridharan et al. (1986) to illus- trate this point. It also was shown by Sridharan et al. (1986) that the zero swell pressure is a function of the dry unit weight of soil, but not of the initial moisture content (Figure 13.16). er funder a pressure of o, + 07 (see Figure 13.15) Table 13.4 Comparison of Zero Swell Pressure Obtained from Conventional Consolidation Tests and Constant Volume Tests—Summary of Test Results of Sridharan et al. (1986) au (KNIT?) Liquid Plasticity Initial void Consolidation Constant Soil Timit index ratio, e, ‘test volume test BC-1 80 4 0.893 294.3 186.4 BC4 98 37 1.002 382.6 251.8 BC-S 96 65 0.742 500.3 3041 BC? 6 66 0372 12753 3728 BCS 4 62 0.656 147.2 687 A Ce Example 13.1 3 A soil profile has an active zone of expansive soil of 6 ft. The liquid limit and the VF" age natural moisture content during the construction season are 50% and 20%, °9P°= tively, Determine the free surface swell.