Soil Science Society of America Journal 64:268-274 (2000)
© 2000 Soil Science Society of America
DIVISION S-5-PEDOLOGY
An Expansive Soil Index for Predicting Shrink–Swell Potential
P.J. Thomasa,
J.C. Bakera and
L.W. Zelaznya
a Dep. of Crop and Soil Environmental Sciences, Virginia Polytechnic Inst. and St. Univ., Blacksburg, VA 24061-0404 USA
pthomas{at}vt.edu
 |
ABSTRACT
|
|---|
Soil properties indicative of shrink–swell potential were studied for 12 soils encompassing several parent materials in Virginia. Soils are rated from moderate to very high shrink–swell potential. The mineralogy classes, soil series, and (parent materials) examined include: smectitic—Iredell (hornblende gneiss), Jackland and Waxpool (diabase); vermiculitic—Kelly (thermal shale); kaolinitic—Cecil (granite gneiss) and Davidson (diabase); and mixed—Carbo and Frederick (limestone), Craven and Peawick (Coastal Plain sediments), and Mayodan and Creedmoor (Triassic sandstones). Soil properties measured were swell index, coefficient of linear extensibility (COLE), particle-size distribution, cation-exchange capacity (CEC), liquid limit, plasticity index (PI), and clay mineralogy. Soils with estimated high or very high shrink–swell potential were clayey, with clay contents exceeding 60%. These expansive soils also exhibited high CEC (>15 cmolc kg-1 soil), high liquid limits (>70), and appreciable swelling 2:1 mineral content (>15% montmorillonite + 1/2 vermiculite on whole-soil basis). An expansive soil rating system, termed the Expansive Soil Index (ESI), was developed using the soil properties most correlated with shrink–swell potential. The sum of swelling 2:1 minerals, swell index, liquid limit, and CEC gave ESI ratings for each soil series. The higher the ESI, the greater the shrink–swell potential. Where less-detailed information is required, such as for initial feasibility studies, an ESI consisting of liquid limit and CEC is recommended. Finally, the soils were grouped into risk categories based on parent material to allow for classification of similar soils into the ESI rating system. Soils with restricted drainage formed from mafic rocks, carbonate parent material, and metamorphic shales are at high risk for expansive soil behavior.
Abbreviations: CEC, cation-exchange capacity • COLE, coefficient of linear extensibility • ESI, Expansive Soil Index • LEP, linear extensibility percentage • PI, plasticity index • PVC, potential volume change • SSA, specific surface area
|
INTRODUCTION
|
|---|
EXPANSIVE SOILS may occur throughout Virginia, but they especially pose a problem where rapid urbanization and development are occurring. As development extends into these areas, identification and quantification of the soil properties that define shrink–swell potential are essential to evaluate properly the stability of a soil as a foundation material.
Shrink–swell classes for each horizon and for the soil profile are based on the change in length of an unconfined clod as moisture content is decreased from a moist to a dry state. If this change is expressed as a percentage, the value used is linear extensibility percentage (LEP). If it is expressed as a fraction, the value used is COLE (Soil Survey Staff, 1996). The shrink–swell classes are defined as follows: low (LEP <3%, COLE <0.03); moderate (LEP 3–6%, COLE 0.03–0.06); high (6–9%, COLE 0.06–0.09); and very high (LEP >9%, COLE >0.09) (Soil Survey Staff, 1997a). Occasionally shrink–swell classes are estimated from accessory soil characteristics such as field-determined plasticity and stickiness and texture. Thus, accurate quantitative measures of linear extensibility are not always available.
Presently, no one method of soil analysis estimates shrink–swell potential accurately for all soils. Soil scientists recognize that shrink–swell behavior can best be predicted by examining a combination of physical, chemical, and mineralogical soil properties. A protocol that integrates these properties and then establishes a shrink–swell model that can be extrapolated across the same or similar parent materials is needed.
Most studies examining expansive soils have been conducted on Vertisols and high base, montmorillonitic (smectitic) soils. In these studies several parameters have been identified as correlated with swelling. Potential volume change of expansive soils in the western USA had been linked to clay content and PI (Holtz and Gibbs, 1956). The variation in swelling of Hapludalfs in Ohio was related to clay content in a study where all other parameters were held constant (McCormack and Wilding, 1975). Swell potentials of montmorillonitic soils in southern Ontario were correlated with clay content and specific surface area (SSA), and SSA explained more of the variability in shrink–swell potential than did clay content (Ross, 1978). In Usterts and Torrerts of arid regions, swell potential, as measured by the COLE, was related to the fine clay content and exchangeable Na percentage (Anderson et al., 1973). Schafer and Singer (1976) determined that the percentage of expandable clays explained most of the variability in swelling potential in soils of Yolo County, California. Shrink–swell potential in kaolinitic and mixed mineralogy soils and acid montmorillonitic soils are often more difficult to predict. Franzmeier and Ross (1968) observed that soils having equal amounts of kaolinite and montmorillonite behaved like montmorillonitic soils, whereas soils with appreciable amounts of montmorillonite had wide ranges in swelling potentials. They postulated the variability was related to the amount of clay and the soil fabric. Acid montmorillonitic and mixed mineralogy Alfisols and Ultisols in Alabama showed weak correlations between COLE and potential volume change (PVC) (Karathanasis and Hajek, 1985). Higher Al saturations may contribute to resistance of the clay minerals to dehydration. Acid conditions were found to favor interlayer formation with Al and Fe in montmorillonite (Carstea et al., 1970) and to inhibit swelling (Rich, 1968). Iron coatings have also been found to reduce swell potential (Davidson and Page, 1956). In a study of Alabama soils ranging from kaolinitic Davidson (Kandiudult) to montmorillonitic Houston soils (Hapludert), CEC was highly correlated with SSA, -1500 kPa moisture content, and PI (Gill and Reaves, 1957).
As can be surmised from the discussion above, several physical, chemical, and mineralogical soil properties influence shrink–swell behavior, with no one property accurately predicting shrink–swell potential for all soils. Often most expansive soils are clayey with high CECs, high SSAs, and high liquid limits. Smectite typically comprises most of the soil clay fraction.
Our study was undertaken with the hypothesis that no one soil property or expansive soil test can precisely predict shrink–swell potential for all soils. However, a set of soil properties that estimates shrink–swell behavior can usually be determined for clayey soils with kaolinitic, mixed, or smectitic mineralogy formed from a variety of parent materials. The objectives of our study were (i) to quantify properties and shrink–swell indices of 12 expansive soils in four major physiographic provinces in Virginia, (ii) to correlate shrink–swell potential with soil properties and shrink–swell indices, (iii) to develop an expansive soil rating system using soil properties correlated with shrink–swell potential to evaluate each soil's propensity to be expansive, and (iv) to develop shrink–swell risk categories for soils within different parent materials.
|
Materials and methods
|
|---|
Study Design
Twelve map units in four physiographic provinces in Virginia comprised the study. These clayey soils have estimated moderate to very high shrink–swell potential with varying amounts of expanding 2:1's, interlayered 2:1's, mica, and kaolinite. Estimated shrink–swell potential for each of the soils for initial placement into expansive soil classes was obtained from the USDA-NRCS database (Table 1)
(Soil Survey Staff, 1997b). All soils are classified (Soil Survey Staff, 1994) as fine, with mineralogy classes encompassing kaolinitic, vermiculitic, smectitic, and mixed families (Table 1).
Map units in this study were delineated in accordance with procedures of the National Cooperative Soil Survey (Soil Survey Staff, 1993). Three delineations in each of the 12 map units were selected from soil survey maps. One pedon within each delineation was excavated to 1.8 m for a total of three pedons per map unit. The most representative pit face was described and sampled. Data from the mid-argillic horizon are presented here since this is the depth at which foundations are typically installed (0.5–0.9 m).
Laboratory Analysis
Samples were air dried, ground, and sieved to remove coarse fragments >2 mm. Laboratory analyses include particle-size distribution, CEC, Atterberg limits, PVC, and clay mineralogy. Particle-size distribution was accomplished by the pipette method (Gee and Bauder, 1986) and CEC by the sum of cations method (NH4OAc, pH 7 and BaCl2-TEA, pH 8.2) (Thomas, 1982). Atterberg limits (liquid limit, PI) were measured by ASTM method D4318 (American Society for Testing and Materials, 1993). Coefficient of linear extensibility was determined by the method of the National Survey Center (Soil Survey Staff, 1996). Potential volume change was determined by the method of Lambe (1960). Shrink–swell potential was determined on each sample on the basis of PVC data. Shrink–swell potential classes are low (<81 kPa), moderate (81–153 kPa), high (153–225 kPa), and very high (>225 kPa) (Soil Survey Staff, 1993). Mineralogical composition was determined by x-ray diffraction and thermal methods. Free Fe oxides were removed with dithionate-citrate-bicarbonate (Mehra and Jackson, 1960). Sand was removed by sieving, and the clay fraction was separated from silts by centrifugation (Jackson et al., 1950). Oriented mounts of the clay fraction were prepared by the method of Rich (1969) and saturated with KCl and MgCl2-glycerol (Whittig and Allardice, 1986). Clay minerals were identified with a Scintag XDS 2000 x-ray diffractometer (Scintag, Santa Clara, CA) with Cu-K
radiation. Thermal analysis of the clay fraction was accomplished with a Dupont 990 Differential Scanning Calorimeter (Dupont, Wilmington, DE). Quantitative estimates of kaolinite and gibbsite were obtained by measuring endothermic peak areas. Quantitative estimates, to the nearest 5%, of other clay fractions were determined by proportioning integrated peak areas of x-ray diffractograms, using kaolinite as an internal standard. Swelling 2:1 minerals were estimated by summing the smectite content and one-half of the vermiculite content, since expansion is limited to two water layers for vermiculite.
Statistical Analysis
Significantly different means of Bt horizon soil properties were separated by least significant difference at the 0.05 level. The Pearson product moment correlation coefficient between variables was used to examine the relationships between soil properties.
|
Results and discussion
|
|---|
Soil Properties
Average values for soil properties for each map unit are presented in Fig. 1 through 7
, and all data for each profile are given in Table 2
. The largest (most extreme) variance within individual map units was observed in clay content and liquid limit (Table 2). The low clay contents and liquid limits corresponded with low values for other measured soil properties. Most of the following discussion refers to average values for soil properties for each of the map units (Fig. 1–7). Shrink–swell classes indicated in the figures (moderate, high, very high) are the USDA-NRCS shrink–swell potential classes (Soil Survey Staff, 1997b).
View larger version (56K):
[in this window]
[in a new window]
|
Fig. 1 Relationship between clay content and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (44K):
[in this window]
[in a new window]
|
Fig. 2 Relationship between cation-exchange capacity (CEC) and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (44K):
[in this window]
[in a new window]
|
Fig. 3 Relationship between liquid limit and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (43K):
[in this window]
[in a new window]
|
Fig. 4 Relationship between plasticity index and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 5 Relationship between COLE and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (49K):
[in this window]
[in a new window]
|
Fig. 6 Relationship between swell index and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 7 Relationship between swelling 2:1 mineral content and estimated shrink–swell class. Means as indicated by the same letter are not significantly different at the 0.05 level of probability
|
|
View this table:
[in this window]
[in a new window]
|
Table 2 Physical, chemical, and mineralogical data of control sections (Bt horizons) and (ESIs) for each profile in the map units
|
|
The Bt horizons of all 12 soils were clayey with clay contents ranging from 363 to 693 g kg-1 (Fig. 1) and 182 to 790 g kg-1 for individual profiles (Table 2). Estimated high and very high shrink–swell soils had similar clay contents, except Creedmoor, which had a much lower clay content (Fig. 1). Moderate shrink–swell soils, Cecil and Davidson, also had high clay contents similar to the more expansive soils. Thus, no apparent relationship between clay content and susceptibility to shrink–swell behavior was observed.
A partial trend in shrink–swell behavior can be observed with CEC. High and very high shrink–swell soils generally had higher CECs than the moderate soils (Fig. 2). However, the Frederick map unit had significantly lower CEC and behaved or belonged in the same class as the moderate shrink–swell soils if only CEC is used to estimate shrink–swell potential. Likewise, Mayodan and Craven, which occur in the moderate to high shrink–swell class, had significantly higher CEC, comparable with Creedmoor, Peawick, Kelly, and Iredell in the high and very high classes.
Liquid limit is the highest in the very high shrink–swell class, intermediate in the high class, and lowest in the moderate class (Fig. 3). Liquid limit demonstrated good correlation with expected shrink–swell class, yet differentiating the moderate from the high shrink–swell class and the high from the very high class was difficult.
Plasticity index, used widely by the geotechnical community to assess shrink–swell potential, showed little correlation with expected shrink–swell class (Fig. 4). However, two distinct groups, which overlap the predefined classes, were indicated. Kelly, Iredell, Jackland, Waxpool, and Carbo had much higher PIs than the remainder of the soils.
Coefficient of linear extensibility and swell index are direct measurements of shrink–swell potential (Fig. 5 and 6). No discernable relationship was observed between estimated shrink–swell class and COLE for the 12 soils (Fig. 5). However, as with plasticity index, two distinct groups separating soils into moderate or high (Cecil, Davidson, Craven, Creedmoor, Peawick, Frederick) and high or very high (Kelly, Iredell, Jackland, Waxpool, Carbo) were observed. The other direct indicator of shrink–swell potential, swell index, showed a high correlation with estimated shrink–swell class (Fig. 6). All four moderate class soils had moderate swell indices (81–153 kPa). Cecil and Mayodan, although having measured swell indices of moderate, bordered the low class (<81 kPa). All four very high soils had average swell indices that placed the soils in the high shrink–swell class (153–225 kPa), although Jackland, Waxpool, and Carbo were borderline to the very high class (>225 kPa).
Swelling 2:1's (smectite, 1/2 vermiculite) had the highest correlation with shrink–swell class, as expected (Fig. 7). Smectitic Waxpool, Jackland, and Iredell had high smectite contents, as does the mixed mineralogy Carbo. The kaolinitic Cecil and Davidson soils had the lowest smectite contents, whereas soils with mixed mineralogy, in both the moderate and high classes, had intermediate smectite contents. The Creedmoor soil averaged 363 g kg-1 clay but had similar smectite content, on a whole-soil basis, to the high clay Frederick. Both had similar swell indices of 
120–150 kPa, further supporting the use of whole-soil smectite content rather than clay content when estimating shrink–swell behavior.
Correlation of Shrink–Swell Properties
All shrink–swell indices measured were positively correlated with each other. Swelling 2:1's, CEC, and liquid limit (indirect measures of shrink–swell potential) were significantly and positively correlated with swell index, a direct measurement of shrink–swell potential (Table 3)
. Values for COLE and PI were not significantly correlated with shrink–swell properties, although other studies have indicated as such (Anderson et al., 1973; Franzmeier and Ross, 1968; McCormack and Wilding, 1975; Schafer and Singer, 1976). The lack of correlation of COLE and PI in this study may be due to Al interlayering or high Fe oxide coating of the clays inhibiting swelling.
Expansive Soil Index
The absolute values of the four soil indices most correlated with predicted shrink–swell potential in this study were swelling 2:1's, CEC, swell index, and liquid limit. Thus, these soil indices were summed into an Expansive Soil Index (ESI) given by the following equation:
The ESI-1 ratings >500 indicate high and very high shrink–swell potential (Table 4)
and would require special design of foundations, such as adding reinforcing bars to footings or installing moisture barriers, to decrease potential expansive soil damage. An ESI-1 <500 describes soils with moderate to high shrink–swell potential. Special design features are suggested to reduce shrink–swell risk, although the design of such features may not be as extensive as required for foundations constructed on higher shrink–swell potential soils.
Identifying and quantifying swelling 2:1 minerals is difficult, time-consuming, and expensive to routinely measure. Only a few laboratories are equipped to make these types of quantifications. Table 3 showed that high values of swell index, liquid limit, and CEC corresponded with high amounts of swelling 2:1 clay minerals. Thus, an alternative ESI, termed ESI-2, was proposed that used only the absolute values of swell index, liquid limit, and CEC. The ESI-1 and ESI-2 were highly correlated (
) and gave similar shrink–swell risk separations of the soils (Table 4). However, separations were not as distinct as when swelling 2:1's are in the equation. An ESI-2 of 250 did appear to separate soils of high risk from soils of very high risk. Demarcation of moderate risk soils from high risk was not as distinct. Thus, conservative estimates of shrink–swell potential are recommended. Although swell index is an easy parameter to measure, there is little swell index (PVC) data in soil survey databases. However, extensive data on liquid limits and CEC are contained in many soil characterization databases. Thus, a third expansive soil index, ESI-3, can be formulated using only liquid limit and CEC as shrink–swell predictors. Fewer categories of estimated shrink–swell risk would be entailed when using this index.
We now have developed three ESIs, each requiring a different level of data input and applying a different level of shrink–swell predictability. What ESI rating should be employed for various intensities of site assessment? We suggest the following guidelines:
ESI-3 — Liquid Limit and Cation-Exchange Capacity
Employed when general information is needed, such as performing feasibility studies for a proposed subdivision or highway, ESI-3 would be sufficient to screen suitable areas from unsuitable areas. Liquid limit and CEC are indirect indicators of shrink–swell potential.
ESI-2 — Swell Index, Liquid Limit, and Cation-Exchange Capacity
Used when site-specific information is needed, application of ESI-2 would include suitability of a site for home foundations or on-site wastewater disposal. Swell index is a direct measure of swelling pressure.
ESI-1 — Swelling 2:1's, Swell Index, Liquid Limit, and Cation-Exchange Capacity
The ESI-1 rating would be used when data is needed in litigation court cases or when additional information is required for foundations or other structures designed to reduce potential damage from shrink–swell soils.
Parent Material Correlation with Shrink–Swell Indices
Many other soils are formed from the same or similar parent materials as the soils described in this study. Thus, it is possible to extrapolate these data and ESI ratings for similar soils on the basis of parent material. Soils with restricted drainage formed from mafic rocks, soils formed from metamorphic shales, and shallow soils formed from carbonate parent material are at very high risk for expansive soil behavior (Table 5)
. Moderately high to high risk was assumed in soils formed from metamorphic shales, soils derived from Triassic sandstones and shales, deep soils formed from carbonate rocks, and Coastal Plain clayey fluvial and marine sediments. Moderate risk can be correlated with soils formed from well-drained felsic and mafic parent materials.
|
Conclusions
|
|---|
Expansive soils cause billions of dollars of damage to homes and property in the USA each year. Damage can be avoided or mitigated if the propensity of a soil to shrink and swell is known before construction. Predicting shrink–swell potential accurately requires both the knowledge of which soil properties influence shrinking and swelling and the magnitude of these parameters. Recognizing the need for quantitative soil information, we developed an expansive soil rating system to assess shrink–swell potential of 12 clayey soils formed from major parent materials that occur throughout Virginia. Expansive Soil Indices are obtained by summing the absolute values of swelling 2:1 minerals, swell index, liquid limit, and CEC. Three levels of precision are available, with highest precision obtained with all four properties. Intermediate precision is available by using swell index, liquid limit, and CEC. Lower precision, but rapid assessment, can be achieved by using only liquid limit and CEC. The ESI system has the flexibility of allowing for the classification of other soils formed from similar parent materials, enabling in rapid, quantitative assessment of shrink–swell potential of a soil.
Received for publication June 18, 1998.
|
REFERENCES
|
|---|
- American Society for Testing and Materials. 1993. Annual book of ASTM standards. Construction. Section 4. Soil and rock; dimension stone; geosynthesis. Vol. 04.08. ASTM, Philadelphia, PA.
- Anderson J.U., Elfadil Fadul Kamal, O'Connor G.A. Factors affecting the coefficient of linear extensibility in Vertisols. Soil Sci. Soc. Am. Proc. 1973;37:298-299.
- Carstea D.D., Harward M.E., Knox E.G. Comparison of iron and aluminum hydroxy interlayers in montmorillonite and vermiculite. I. Formation. Soil Sci. Soc. Am. Proc. 1970;34:517-521.
- Davidson S.E., Page J.B. Factors influencing swelling and shrinking in soils. Soil Sci. Soc. Am. Proc. 1956;20:320-324.
- Franzmeier D.P., Ross S.J., Jr. Soil swelling: Laboratory measurement and relation to other soil properties. Soil Sci. Soc. Am. Proc. 1968;32:573-577.
- Gee G.W., Bauder J.W. Particle-size analysis. In: Klute A., ed. Methods of soil analysis, Part 1, 2nd ed Madison, WI: ASA and SSSA, 1986:383-411 Agron. Monogr. 9..
- Gill W.R., Reaves C.A. Relationships of Atterberg limits and cation-exchange capacity to some physical properties of soil. Soil Sci. Soc. Am. Proc. 1957;21:491-494.
- Holtz W.G., Gibbs H.J. Engineering properties of expansive clays. Trans. ASCE 1956;121:641-677.
- Jackson M.L., Whittig L.D., Pennington R.P. Segregration procedure for the mineralogical analysis of soils. Soil Sci. Soc. Am. Proc. 1950;14:77-81.
- Karathanasis A.D., Hajek B.F. Shrink–swell potential of montmorillonitic soils in udic moisture regimes. Soil Sci. Soc. Am. J. 1985;49:159-166.[Abstract/Free Full Text]
- Lambe T.W. The character and identification of expansive soils. Washington, DC: U.S. Gov. Print. Office, 1960 Fed. Housing Admin. Rep. 701..
- McCormack D.E., Wilding L.P. Soil properties influencing swelling in Canfield and Geeburg soils. Soil Sci. Soc. Am. Proc. 1975;39:496-502.
- Mehra O.P., Jackson M.L. Iron oxide removal from soils and clays by a dithionite-citrate system buffered with sodium bicarbonate. Clays Clay Min. 1960;7:317-327.
- Rich C.I. Hydroxy interlayers in expansible layer silicates. Clays Clay Min. 1968;16:15-30.[ISI]
- Rich C.I. Suction apparatus for mounting clay specimens on ceramic tile for x-ray diffraction. Soil Sci. Soc. Am. Proc. 1969;33:815-816.
- Ross G.J. Relationships of specific surface area and clay content to shrink–swell potential of soils having different clay mineralogical compositions. Can. J. Soil Sci. 1978;58:159-166.
- Schafer W.M., Singer M.J. Influence of physical and mineralogical properties on swelling of soils in Yolo County, California. Soil Sci. Soc. Am. J. 1976;40:557-562.[Abstract/Free Full Text]
- Soil Survey Staff. 1993. Soil survey manual. USDA-SCS Agric. Handb. 18. U.S. Gov. Print. Office, Washington, DC.
- Soil Survey Staff. 1994. Keys to soil taxonomy. 6th ed. Pocahontas Press, Blacksburg, VA.
- Soil Survey Staff. 1996. Soil survey laboratory methods manual. Soil Survey Investigations Rep. 42. Ver. 3. National Soil Survey Center, Lincoln, NE.
- Soil Survey Staff. 1997a. Map unit interpretation records database. Available at http://www.statlab.iastate.edu/soils/muir/ (verified 22 Nov. 1999).
- Soil Survey Staff. 1997b. National soil survey handbook. USDA-NRCS Publ. 430-VI. U.S. Gov. Print. Office, Washington, DC.
- Thomas G.W. Exchangeable cations. In: Page A.L., ed. Methods of soil analysis, Part 2, 2nd ed Madison, WI: ASA and SSSA, 1982:159-165 Agron. Monogr. 9..
- Whittig L.D., Allardice W.R. X-ray diffraction techniques. In: Klute A., ed. Methods of soil analysis, Part 1, 2nd ed Madison, WI: ASA and SSSA, 1986:331-362 Agron. Monogr. 9..
TOP
ABSTRACT
INTRODUCTION
