Cone Index of Loamy Sands as Influenced by Pore Size Distribution and Effective Stress1

M. J. VEPRASKAS2

ABSTRACTTillage pans form in coarse textured soil horizons of the Atlantic

Coastal Plain and develop high cone index values (soil strengths)which slow root growth and may decrease crop yields. Identificationof soil properties that influence cone index may improve field iden-tification of tillage pans and lead to the development of techniquesto reduce their detrimental effects. This study examined the influ-ence of pore size distribution, and effective stress on the cone indexof loamy sand and loamy fine sand soil materials. Effective stressvalues were calculated from soil water characteristic curves and com-pared to cone index values for both natural cores and remolded ma-terials. The remodeled samples of a loamy sand and a loamy finesand developed different cone index values for relative saturationsbetween 1.00 and 0.10, and these cone index differences were par-tially related to pore size distribution differences between materials.Pore size distribution influenced cone index by controlling effectivestress. In both remolded samples and undisturbed cores of loamysand, effective stress values ranged from 0 to 33 kPa for soil waterpotentials between 0.0 and —1500 kPa, respectively. For relativesaturation values between 1.00 and 0.10, changes in effective stressclosely conformed to changes in cone index for nondispersed soilmaterials. At a given effective stress, cone index was influenced bybulk density and soil particle characteristics that influence inter-particle friction.

Additional Index Words: Atlantic Coastal Plain soils, bulk den-sity, compaction, soil strength, tillage pan.

Vepraskas, M.J. 1984. Cone index of loamy sands as influenced bypore size distribution and effective stress. Soil Sci. Soc. Am. J.48:1220-1225.

MANY loamy sand and sandy loam Ap and E ho-rizons in the Atlantic Coastal Plain region have

a tillage pan. The tillage pans have few macroporesand develop soil strengths sufficiently high to restrictrooting depth and decrease crop yields (Campbell etal., 1974). At present, a deep tillage method such assubsoiling is the most common method used to dis-rupt tillage pans. Identification of soil properties thatexert a great influence on soil strength may lead tomeans of lowering detrimental strengths through man-agement practices in addition to tillage. Such man-agement practices could include use of soil condition-ers to alter aggregation and pore size distribution.

Numerous investigations have shown that watercontent, bulk density, and texture influence soilstrength and cone index (penetrometer resistance)(Barley et al., 1965; Camp and Gill, 1969; Gerard,1965; Jamison and Weaver, 1952; Taylor and Gard-ner, 1963; Taylor et al., 1966). Pore size has also beenrelated to soil strength in several studies. For example,Byrd and Cassel (1980) examined cone index valuesof reconstituted soil materials containing 66.6 to 82.2%sand, all with a bulk density of 1.79 ± 0.03 Mg m~3.

1 Contribution from the Dep. of Soil Science, North Carolina StateUniversity. Paper no. 9414 of the Journal Series of the North Car-olina Agricultural Research Service, Raleigh, NC 27695-7601. Re-ceived 13 Oct. 1983. Approved 25 June 1984.2 Assistant Professor of Soil Science, North Carolina State Univ.,Raleigh, NC 27695-7619.

Regression equations were developed relating cone in-dex to water content, percent sand, and volume ofpores with diameters > 160 pm. Gerard et al. (1966)examined the relationship between soil strength (mod-ulus of rupture) and pore size for remolded materialshaving sand contents of 55 and 70%. They found soilstrength decreased as the volume of voids with di-ameters > 50 Mm increased.

Stitt et al. (1982) examined the influence of 20 soilphysical, chemical, and mineralogical properties oncone index for 50 tillage pan materials from the NorthCarolina Coastal Plain. They found that water con-tent, bulk density, the dense soil angle of repose, andthe volume of pores with neck diameters between 74and 136 Mm accounted for 75% of the variation in theobserved cone index values. No simple relationshipbetween texture and cone index was found, nor wasthere any evidence that cementing agents were affect-ing the cone index values.

On the basis of these studies, it appears that poresize is related to soil strength, although its influencehas not been studied in detail. Its mode of action onsoil strength is not known but may be interrelated tothe effects of bulk density, water content, and particlesize distribution on soil strength. The primary purposeof this investigation was to examine the influence ofpore size distribution on the cone index values ofloamy sand soil materials. In addition, the relation-ships of water content, bulk density, and particle sizedistribution to cone index were also evaluated.

MATERIALS AND METHODSThree separate experiments were conducted to evaluate

the effects of water content, bulk density, particle size dis-tribution, and pore size distribution on the cone index ofcoarse-textured materials subject to tillage pan formation.Experiment I

The effects of water content and bulk density on the coneindex of two loamy sand soil materials were examined. Fiftykilograms of E horizon material were collected from twolocations on the Atlantic Coastal Plain. A Wagram loamyfine sand (loamy, siliceous, thermic, Arenic Paleudults) wassampled near Whiteville, NC, and a Wagram loamy sandwas sampled near Clayton, NC. Samples were taken frompits adjacent to those sampled by Cruse et al. (1980), whofound that these soils differed in soil particle shape, particlesize distribution, and bulk density.

All soil materials were air-dried and passed through a 250-nm sieve to reduce aggregate size. This was required toachieve specific bulk densities by vibration. Materials >250Mm and <250 pm in diameter were recombined after siev-ing. Oven-dried (105°C) materials were vibrated for 30 sinto aluminum cups (6.3 cm in diameter by 4.3 cm in height)using the vibrator method described by Cruse et al. (1980).Three bulk density classes were achieved for each soil byvarying vibrator voltage from 20 to 60 V. Remolded bulkdensity values ranged from 1.57 to 1.80 Mg m~3 for theloamy sand, and 1.41 to 1.58 Mg m~3 for the loamy finesand. Cruse et al. (1980) reported similar results.

The packed soil cups were covered with cheesecloth, in-verted, and placed on ceramic desorption plates, and satu-

1220

VEPRASKAS: CONE INDEX OF LOAMY SANDS 1221

Table 1—Particle size distributions._______Sand, mm_______ Silt, jim Clay,

2.0- 1.0- 0.5- 0.25- 0.1- Total 50- 20- Total ——Soil mix 1.0 0.5 0.25 0.1 0.05 sand 20 2 silt <2

Loamy fine sandWhole soilSand and siltSand

0.30.8trf

0.70.90.7

2.02.22.6

67.067.675.1

17.618.920.4

87.690.498.8

4.67.00.8

3.62.20.4

8.29.21.2

4.20.4tr

Loamy sandWhole soilSand and siltSand

2.72.32.6

17.420.821.8

24.226.328.5

33.734.637.2

8.67.79.1

86.691.799.2

3.94.70.8

7.83.2tr

11.77.90.8

1.70.4tr

t Indicates percentage was < 0.1%.

rated overnight. Four replicated cores for each bulk densitywere equilibrated in pressure chambers at soil water poten-tials (f s) of 0, -10, -33, -100, -500, and -1500 kPa.Cone index was determined on each sample with a hand-cranked cone penetrometer using a 30° cone with a 6-mmbasal diameter. The penetrometer tip was advanced into eachcup at a rate of 2 mm s~' and the force (kg) was measuredusing a dial gauge and proving ring at a depth of 2 cm whichwas approximately the center of the cup. Only one forcemeasurement was made for each sample. Cone index wascalculated as force (kg) divided by the basal area of the cone.All samples were then oven-dried (105°C) for 12 h, and watercontent determined gravimetrically.

Experiment IIThe effects of clay and silt contents on cone index were

analyzed for the whole soil materials (<2 mm diam), mix-tures of sand and silt separates (2-0.002 mm diam), andsands only (2-0.05 mm diam). A sand and silt mix wasprepared from the loamy sand material by placing 2 kg ofsoil in a bucket, adding 7 L of 0.2M Na3PO4, stirring, andsiphoning off clay (<2 /tm) after sands and silts settled. Thisprocess was repeated until the supernatant was clear. Thesand mix was prepared similarly but with both clays andsilts siphoned off. Similar mixes were prepared from theloamy fine sand as well. Each mix was oven-dried (105°C),vibrated (60 V for 30 s) into Al cups, and saturated. Fourreplicates were desorbed to each of the following '̂s: — 1.0,-2.5, -5.0, -7.5, -10, -33, -100, -500, and -1500kPa. Soil water characteristic curves were determined foreach mix. Cone index values and gravimetric water contentswere determined as described previously.

Particle size distributions of the loamy fine sand and loamysand materials and the mixes derived from them were de-termined by the pipette method and dry-sieving of sands.Results are shown in Table 1.

Experiment HIUndisturbed soil cores were collected from the Ap and E

horizons of a Wagram loamy sand near Clayton, NC to de-termine relationships among pore size distribution, effectivestress, and cone index. Samples were collected from a fieldthat was bedded and planted in tobacco (Nicotiana tabacumL.). Samples were collected in September after the tobaccowas harvested. All samples were collected in 7.6 cm by 7.6cm metal rings between adjacent plant stalks at three posi-tions measured from the Ap-E horizon boundary. Sampleswere collected between 0 to 15 cm and 15 to 30 cm abovethe Ap-E horizon boundary, and 0 to 15 cm below the Ap-E horizon boundary. Thirty-three cores were collected fromeach depth along three adjacent crop rows.

Samples were air-dried for 180 d, and then oven-dried(105°C) to determine bulk density. Samples were then di-vided into three groups on the basis of bulk density; the

bulk density mean and standard deviation for each groupwere: 1.40 ± 0.04 (Ap horizon samples only), 1.66 ± 0.03(Ap and E horizon samples) and 1.79 ± 0.02 (E horizonsamples only). All samples were covered with cheesecloth,placed on ceramic desorption plates, and saturated over-night. For each bulk density group, four cores were equili-brated to one of the following i '̂s: —2.5, —5.0, —7.5, —10.0,-33, -100, and -1500 kPa. After equilibrating for 48 h,the cores were weighed, and the cone index determined asdescribed previously with the exception that three forcemeasurements were made per core at 2-cm depth incrementsand these were averaged for the entire core. All cores werethen oven-dried (105°C) and the water contents determinedgravimetrically. Soil water characteristic curves were devel-oped for each of the three bulk density groups.

RESULTS AND DISCUSSIONResults of experiments are presented by first ex-

amining relationships among cone index (CI), bulkdensity (Db), and gravimetric water content (w). Em-phasis is given to the shapes of the CI-w curves whichwill be shown to be related to the shapes of the watercharacteristic curves and to soil pore size distribution,and also to changes in effective stress. Finally, CI willbe related to effective stress, and the influence of in-terparticle friction on this relationship will be dis-cussed.

Relationships among Cone Index, Bulk Density, andWater Content

Relationships among CI, Db, and w for remoldedWagram loamy fine sand (LFS) and Wagram loamysand (LS) materials are shown in Fig. 1. For both ma-terials the highest CI values measured occurred be-tween w's of 0.02 to 0.03 kg kg"1 for each Db exam-ined. These water contents correspond to a $ of — 1500kPa. At w's >0.08 kg kg-', the LFS CI curves wereflatter in shape than the LS CI curves. For both ma-terials, the CI curve shapes did not appear to be af-fected by changes in Db.

Cone index values as a function of relative satura-tion (S) are shown in Fig. 2 for the LFS and LS ma-terials, and the sand and silt, and sand mixes derivedfrom them. Curves of best fit were drawn to connectthe data points. Bulk density values for the LFS ma-terials were similar for each mix (Fig. 2A). Cone indexcurve shapes were similar among the LFS mixes forS*s >0.5, with the LFS having the lowest CI values.For S"s <0.5, CI curve shapes differed among the LFSderived mixes. Cone index values generally increasedwith decreasing S for the LFS and the sand and siltmix, but tended to decrease for the sand.

Bulk density values differed among LS mixes withthe sand and silt mix having the highest Db and thehighest CI values at 5"s <0.8 (Fig. 2B). The CI curveshapes of the sand and silt and sand mixes differedfrom that of the LS material, particularly at S values> 0.3. However, no CI values could be obtained forS values between 0.5 to 0.7 (indicated by the dashedline in Fig. 2B), because too much water drained fromthe pores as i^'s were increased from —2.5 to —5.0kPa to make CI measurements at intermediate S val-ues obtainable. Thus, the actual differences in CI curveshapes between the LS and other mixes can only beestimated for 5°s between 0.5 and 0.7. The sand mix

1222 SOIL SCI. SOC. AM. J., VOL. 48, 1984

BULK DENSITY(Mg/m3)

1.5740.01 A A1.45*0.01 o—O1.41*0.01 •__•

A.

.04 .08 .12 .16 .20WATER CONTENT (kg/kg)

.24 .28

BULK DENSITY

1.8010.01 * —— *1.67 ±0.01 O — O158*0.01 • —— •

.04 .08 .12 .16 .20 .24 .28WATER CONTENT (kg/kg)

Fig. 1—Cone index values as functions of water content and bulkdensity for remolded loamy fine sand (A) and loamy sand (B)whole soil materials. Curves were drawn to connect groups of datapoints.

had the lowest CI values at S°s between 0.10 and 0.16,as was found for the LFS-derived materials. The sandfractions derived from both the LFS and LS materialshad similar CI values and curve shapes despite Dbdifferences of 0.20 Mg rn~3.

Relationships of Pore Size Distribution to EffectiveStress and Cone Index

The soil water characteristic curves for the LFS, LS,and related materials are shown in Fig. 3. Shapes ofthe water characteristic curves were similar to theshapes of the CI curves of Fig. 2. Large slope changesgenerally occurred in both the CI curves and watercharacteristic curves at similar S values, primarily be-tween 0.7 to 0.9, and 0.1 to 0.3. If it is assumed thatthe soil water characteristic curves are a function ofthe pore size distribution, then the CI curves wererelated to pore size distribution and to the interactionbetween soil water content and \}/. Water content and\l/ influence soil strength by controlling effective stress(Williams and Shaykewich, 1970; Towner and Childs,1972). In simplest terms, the effective stresses for theunsaturated soil materials studied here are equal tothe pressures with which soil particles are pulled to-gether by water films. Effective stress values for each

BULK DENSITY(Mg/m3)

I02

I04J

X I03Iui

o

I02;

•——• I.58±O.OI

O——O 1.60*0.01

1.57 ±0.01

0.2 0.4 0.6 0.8RELATIVE SATURATION

MIXBULK DENSITY

(Mg/m3)

LOAMY SAND <

SAND ANDSILT

SAND

1.70 + 0.01

1.87 + 0.02

X——x 1.78 ±0.01

0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

Fig. 2—Cone index values as functions of water content and particlesize distribution for loamy fine sand (A) and loamy sand (B) ma-terials and the mixes derived from them by removal of the missingconstituents.

of the six soil mixes studied were calculated from thesoil water characteristic curves (Fig. 3) using the equa-tion of Towner and Childs (1972):

a = [1]

where a' is effective stress, pSd is the prevailing soilwater potential, S is the relative saturation, a is theproportion of the water in a saturated soil that can bedrained by a suction (in this case a ^ of — 1500 kPa),and sd is a soil water potential. Effective stress valuesare shown in Fig. 4 for each soil mix. The integral inEq. [1] was evaluated graphically.

Shapes of the a' curves were similar to the shapesof the CI curves in Fig. 2. Changes in CI with decreas-ing S1 generally conformed to changes in a'. The mostsignificant departure of CI values from the a' curvesoccurred in the sand and silt mixes for 5°s <0.5. Ingeneral, CI values of the sand and silt mixes increasedas S decreased from 0.5 to 0.2, while a' decreased overthis moisture range.

Effective stresses were also calculated for undis-turbed soil cores of Wagram loamy sand for three Db's(Fig. 5). Effective stress values of the undisturbed andremolded materials were similar (Fig. 4 and 5). Thecalculated a' curves of the undisturbed cores are com-pared to measured CI values in Fig. 6 where goodagreement was found between changes in CI and theshape of the a' curves for S values between 1.00 to0.10.

VEPRASKAS: CONE INDEX OF LOAMY SANDS 1223

A. 40 ^

0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

0.2 0.4 0.6 0.8RELATIVE SATURATION

1.0

Fig. 3—Soil water characteristic curves for loamy fine sand (A) andloamy sand (B) materials and the mixes derived from them byremoval of the missing constituents.

The a' values computed from Eq. [1] consist of twocomponents determined by each term of the equation.The first term Lsrf [1 — (1 —S)/«]} represents that a'component produced by the prevailing \f/ in water-filledpores. The second term is the a' component producedby isolated water films which may form annular ringsat particle contact points (Towner and Childs, 1972).This water remains in a soil after a pore has drained.The value of each term as a function of S is shown inFig. 7 for the undisturbed cores. At S values >0.4, a'is dominated by the stress produced by the prevailing\l/ in water-filled pores. The stress produced by isolatedwater films in drained pores has its greatest influenceat S values <0.3.

Byrd and Cassel (1980) found that CI was negativelycorrelated with the volume of pores with diameters>160 tan. Water held in these pores drains at i/-'s be-tween 0 and —1.8 kPa. The results discussed hereinsuggest that because pores having diameters > 160 nmdrain at small '̂s, their water is not utilized for de-veloping the effective stress produced by the prevail-ing ^ in water-filled pores. Thus, increasing the vol-ume of pores with diameters > 160 pm can be expectedto decrease CI.

LOAMY FINESAND

SAND ANDSILT

SAND

;40

0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

LOAMY SAND

0 0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

Fig. 4—Effective stress values as functions of water content for loamyfine sand (A) and loamy sand (B) soil mixes. The effective stressvalues were calculated from the water characteristic curves ofFig. 3.

i20

BULK DENSITY(Mg/m3)

o:£Ld

HObj

10:

1.0 £

0 0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

Fig. 5—Effective stress values as functions of water content for un-disturbed cores of loamy sand at three bulk densities.

I04:

I03:

OoI02

0 0.2 0.4 0.6 0.8 I 0RELATIVE SATURATION

Fig. 6—Cone index as a function of water content and bulk densityfor undisturbed cores of loamy sand material. Solid lines are theeffective stress curves from Fig. 5 drawn to show how the curveshapes fit the CI data points.

Relationship between Cone Index and Effective StressRelationships between a' and CI are shown in Fig.

8 for the LFS and LS materials and the mixes derived

1224 SOIL SCI. SOC. AM. J., VOL. 48, 1984

10.0

EFFECTIVESTRESS COMPONENTS

1.40+0.041.66+0.031.79+0.02

0.2 0.4 0.6 0.8 1.0RELATIVE SATURATION

Fig. 7—Stress values for two effective stress components as functionsof water content and bulk density. Stress values were calculatedfrom the water characteristic curves obtained for undisturbed coresof loamy sand.

from them. The CIV relationship was linear for bothwhole soil materials and regression equations areshown for each material. The CIV relationship forthe sands may also be linear, but the available dataare from too narrow a range of CI or a' values to seedefinite relationships. The CIV relationships for thesand and silt mixes were not linear over the entirerange of a' values studied, particularly for the mix de-rived from the LS. As noted earlier, the CI values ofthe sand and silt mixes increased while a' values de-creased for S values between 0.5 and 0.2 (Fig. 2 and4).

The CIV relationships can be interpreted by con-sidering the shear strength of sand. According toMitchell (1976, p. 284) the shear strength of sands asa function of effective stress, when cementing agentsare not present, is given by:

= ay tan<// [2]where r«-is the shear strength along a failure plane, ayis the effective stress normal to the failure plane, and<t>' is a measure of the frictional resistance among grains.Equation [2] is similar to the regression equationsshown in Fig. 8 for the whole soil materials with CIbeing analogous to but not equal to rff. The large neg-ative constant term in the equation for the LS materialresults from the fact that its CI values remained vir-tually constant (ranging from 0.3 to 0.5 kPa) for a'values from 0 to 1.5 kPa. This is attributed to exper-imental error, in part due to some of the CI valueshaving been estimated from the dashed line in Fig.2B. Assuming Eq. [2] is analogous to the CIV regres-sion equations for the LFS and LS materials (Fig. 8),then the slopes of the latter equations are a functionof the frictional resistance among grains. In generalfrictional resistance among grains is affected by par-ticle size distribution, particle shape and roughness,mineralogy, particle arrangements (Lambe and Whit-man, 1979) and bulk density (Williams and Shayke-wich, 1970). Differences in the CIV relationshipsamong the materials shown in Fig. 8 are probably dueto a combination of factors including differences in

8000

£6000

Q4000

2000

• LOAMY FINE SANDO SAND AND SILT

> SAND

,CI=l08(<r)+2r = 0.99»*

eooo

6000

40OO

G 2000

2 4 6 8 10 12 14 16 18 20EFFECTIVE STRESS (kPo)

• LOAMY SAND

O SAND AND SILT

« SAND

Cl=l57(r)- l53r*0.99**

142 4 6 8 10 12EFFECTIVE STRESS (kPa)

Fig. 8—Relations between effective stress and cone index for theloamy fine sand (A) and loamy sand (B) materials and mixesderived from them.

particle size distribution, particle shape and arrange-ment, and bulk density among the materials studied.The distinct nonlinearity of the CIV relationship forthe sand and silt mixes in Fig. 8 is believed to be dueto a progressive movement of the dispersed silt par-ticles to the sand grain contact points on drying. Thistype of fabric rearrangement would increase frictionalresistance among sand grains on drying and cause CIto increase even though a' might decrease or remainconstant as was found for 5"s between 0.1 and 0.5(Fig- 4).

Cone index values determined on natural cores ofLS are compared to their corresponding a' values inFig. 9. The CI values for a particular a' increased with

8000

0 2 4 6 8 10 12 14EFFECTIVE STRESS (kPa)

Fig. 9—Relations between effective stress and cone index for naturalcores of loamy sand at three bulk densities.

HORTON ET AL.: SOIL TEMPERATURE IN A ROW CROP WITH INCOMPLETE SURFACE COVER 1225

increasing Db. As Db increases, particle contacts prob-ably increase resulting in greater interparticle friction.At each Db, the slope of the CIV curves decreased ata' values >4 kPa. The CIV relationship for a' values>4 kPa is tentative because large increases in a' oc-curred over small changes in S1 (Fig. 5), and correla-tions between a' and CI values are probably subjectto large errors for a' values >4 kPa.

Results of these experiments suggest that soil poresize distribution influences CI by controlling changesin effective stress. In nondispersed materials and sands,changes in a' closely conformed to changes in CI for5 values between 1.00 and 0.10. The CI value for aparticular </ is influenced by Db and factors that affectfrictional resistance among particles. It should be notedthat the magnitude of any CI measurement dependson a variety of factors including penetrometer coneangle and basal area, rate of penetration, and the sam-ple size (Cassel, 1982). Therefore, the CI values re-ported in this study would probably not be the samefor a different penetrometer, although the general CIrelationships described would be the same.

ACKNOWLEDGMENTSThe assistance of M. Carpenter, R. Pearson, and F. Av-

erette in performing laboratory analyses for this study isgratefully acknowledged.