Download - Soil compaction by wheeling: changes in soil suction ... · Soil compaction by wheeling: changes in soil suction caused by compression ... oedometrique standard´ ´equip ´ed’untensiom

European Journal of Soil Science, August 2010, 61, 599–608 doi: 10.1111/j.1365-2389.2010.01245.x

Soil compaction by wheeling: changes in soil suctioncaused by compression

K . C u ia∗ , P . D e f o s s e za,b∗∗ , Y . J . C u ic & G . R i c h a r dd

aINRA, Unite Agro-Impact, UR1158, rue F. Christ, 02007 Laon, France, bINRA, UMR614 FARE, 2 Esplanade Roland Garros, BP 224,51686 Reims Cedex 2, France, cEcole des Ponts - ParisTech, UR Navier/CERMES, 6 et 8 avenue Blaise Pascal, Cite Descartes,Champs-sur-Marne, 77455 Marne-la-Vallee Cedex 2, France, and dINRA, UR0272 Science du Sol Orleans, 2163 Avenue de la Pomme dePin, BP 20619 Ardon, 45166 Olivet Cedex, France

Summary

Soil compaction caused by agricultural machinery has been increasingly recognized as a considerable problemfacing intensive agriculture. Most of the models used to estimate soil deformation during the passage ofmachines are based on the concept of total stress: they have neglected an important stress variable forunsaturated soils; that is, the matric suction. The aim of the present work was to evaluate the validity of thishypothesis by studying suction variation during a static compression test. A standard oedometer cell equippedwith a tensiometer was used to measure soil suction in situ for different vertical stresses. Measurementswere carried out on remoulded soil samples obtained by compacting a loamy soil at different initial watersuctions (<100 kPa). The results showed that the suction remained almost constant until a stress thresholdvalue, σt, beyond which the suction decreased as the stress increased. This stress threshold increased with theinitial suction. These results corroborated the hypothesis of a constant suction during deformation, which isusually assumed to model soil compaction during traffic for soils with suction greater than 20 kPa. The resultsobtained highlighted the effect of soil structure on the stress threshold: σt was greater for soil samples withinitial aggregates <2 mm than for those with initial aggregates <0.4 mm. This was interpreted at the porescale by comparing qualitatively the change in pore-size distribution and the expected distribution of water inthe pores. This interpretation was based on pore-size distribution measurement by mercury intrusion.

Compaction de sols par le passage d’engins: changement de succion cause par lacompression

Resume

Le tassement des sols par les engins agricoles est un probleme important auquel est confrontee l’agricultureintensive. La plupart des modeles utilises pour prevoir la deformation d’un sol sous le passage d’un engin estbasee sur le concept de contrainte totale. Ils negligent alors une variable importante pour les sols non satures:la succion du sol. L’objectif de cette etude est d’evaluer les conditions d’application d’une telle hypothese apartir de l’observation des variations de succion lors d’un test de compression en laboratoire. Un dispositifoedometrique standard equipe d’un tensiometre dans la cellule de compression a permis de mesurer in situla succion du sol lors de l’application de differentes contraintes verticales. Les mesures ont ete realisees surun sol de limon tamise et porte a differentes succion initiales (<100 kPa). Les experiences montrent que lasuccion reste pratiquement constante jusqu’a une valeur seuil de contrainte verticale σt au-dela de laquelleelle decroıt a mesure que la contrainte verticale augmente. Ce seuil de contrainte verticale σt augmente avecla succion initiale. Ces resultats corroborent l’hypothese d’une succion constante au cours de la compression

∗Present address: Southwest Jiaotong University, School of Civil Engi-neering, 610031 Chengdu, Sichuan, P. R. China.∗∗Present address: Inra, UR1263 EPHYSE, 71 Avenue Edouard Bourlaux,F-33140 Villenave d’Ornon, France.

Correspondence: P. Defossez. E-mail: [email protected]

Received 29 April 2008; revised version accepted 28 February 2010

© 2010 The AuthorsJournal compilation © 2010 British Society of Soil Science 599

600 K. Cui et al.

utilisee le plus souvent dans les modeles de compactage mais pour des succions superieures a 20 kPa. Lesresultats mettent en evidence un effet de la structure initiale des echantillons : le seuil de contrainte verticaleσt est superieur pour des sols tamises <2 mm par rapport a ceux tamises <0.4 mm. Ceci a ete interprete al’echelle des pores en comparant l’evolution des distributions de taille de pore et la distribution supposee del’eau dans ces pores. Cette interpretation repose sur des mesures de porosimetrie a mercure.

Introduction

Compaction caused by agricultural machines greatly modifies soilstructure in both tilled and underlying layers. Because of itspersistence, compaction of subsoil layers can be considered as along-term degradation but compaction also concerns surface layersbecause it impacts significantly on plant growth and production(for example through effects on root penetration) and environment(through runoff and greenhouse gas emissions) (O’Sullivan &Simota, 1995). Compaction is defined as the deformation processof cultivated soil in which a rut is formed on the surface,thus decreasing bulk porosity, under the wheels of agriculturalmachines. Changes in pore shape caused by shearing also occur.Numerical analyses using the Finite Element Method have beenused to simulate soil compaction on the basis of stress-strainbehaviour and the mechanical parameters involved have generallybeen estimated from laboratory tests. Gysi (2001) modelled thecompaction of a loamy soil under heavy wheel traffic using theModified Cam-Clay model in PLAXIS code (PLAXIS, 1998).Kirby (1994) simulated the deformation of an agricultural clay soilusing a simple extension to the Modified Cam-Clay model. Thesenumerical studies showed that the critical-state models originallydeveloped for saturated soils can be used for unsaturated soilusing constant mechanical parameters measured in unsaturatedconditions. However, some authors (Burland, 1956; Jennings &Burland, 1962) showed that classical soil mechanics for saturatedsoils is unable to explain the mechanical behaviour of unsaturatedsoils satisfactorily. To achieve this it would be necessary toconsider two independent variables: net stress (σ − ua) where σ

is the total stress and ua is the pore-air pressure, and suction s =ua − uw, where uw is the pore-water pressure. In particular, it canbe assumed that for agricultural conditions at small vertical stress(<400 kPa), the air in the pore space connects and ultimatelyinterconnects with the atmosphere. Under these conditions, ua isequal to zero and the two stress variables become the total stress(σ ) and pore-water pressure (uw). Different constitutive modelshave also been proposed for unsaturated soils for geotechnicalpurposes (Alonso et al., 1990; Wheeler & Sivakumar, 1995;Cui & Delage, 1996) and for compaction of agricultural soils(Richards, 1992).

The present work deals with the variation of soil matric suctionunder the application of stress caused by the passage of vehiclesin soils used for agriculture. In comparison with soils consideredfor geotechnical applications, cultivated soils are more porous,especially in topsoil layers. Compaction problems occur in wetconditions, usually for suctions <100 kPa when the loading time

is short (t < 0.1 s) and vertical stresses σ are generally less than400 kPa. Some authors have studied the soil suction of samplesunder different levels of compression stress in conditions relevantfor compaction of agricultural soil (Larson & Gupta, 1980;Wulfsohn et al., 1998; Tarantino & Tombolato, 2005; da Veigaet al., 2007). Soil suction remained quasi-constant or increasedfor compressive stresses smaller than a given stress threshold(Larson & Gupta, 1980; Wulfsohn et al., 1998; da Veiga et al.,2007). This stress threshold was related to the saturation degree ofsoils (Larson & Gupta, 1980). In contrast, Tarantino & Tombolato(2005) studied the change of suction after compaction on clayand reported that suction decreased systematically. Continuoussoil suction monitoring (Peng et al., 2004) during compressionshowed non-equilibrium effects caused by air and water drainageprocesses. Peng et al. (2004) and Krummelbein et al. (2008)showed the effects of loading time on the soil suction change in thetransient regime that follows the stress application. These changeswere related to the precompression stress (Peng et al., 2004).

The present work examines the effects of initial soil suction andinitial soil structure on soil suction changes under static compres-sion after equilibrium, by carrying out oedometer compressiontests with measurements of sample soil suction based on tech-niques developed for unsaturated soils in geotechnical engineering(Dineen & Burland, 1995; Ridley & Burland, 1996; Tarantinoet al., 2000; Tarantino & Mongiovì, 2001, 2002; Ridley et al.,2003). Particular attention was paid to the relationship betweenthe changes in suction and soil properties in terms of initial suc-tion, degree of saturation and pre-compression stress. Changes inpore-size distribution at two stages of loading were also analysedby using the mercury intrusion technique to interpret qualitativelythe different features observed in the mechanical tests.

Materials and methods

For this study, a loamy soil (Hortic luvisol, 173 g clay kg−1,777 g silt kg−1, 50 g sand kg−1) from the INRA experimentalfarm located at Mons in northern France was used. The soilhad a liquid limit of 0.29 g g−1, a plasticity index of 0.06 g g−1

and a solid density of 2.7 Mg m−3. Air-dried soil was passedthrough 2- and 0.4-mm sieves and then stored at a water contentof 0.02 g g−1.

Oedometer compression test with soil suction measurements

Matric suction measurements. The matric suction of soil spec-imens was measured with a tensiometer inserted through an

© 2010 The AuthorsJournal compilation © 2010 British Society of Soil Science, European Journal of Soil Science, 61, 599–608

Compression effects on soil compaction 601

displacementtransducers

load piston

oedometer cell

(a)

load piston

black membrane

load cap

soil specimen

base pedestal

tensiometer

1.5 cm

(b)

Figure 1 Oedometer equipped with a tensiometer for suction monitoring(a). Details of the oedometer cell with a neoprene membrane coveringthe cell to prevent soil water evaporation. An air pocket 15 mm highis provided between the membrane and the cell for air expulsion duringloading (b).

porous stone seal

diaphragm

strain gauge

water reservoir

Araldite

electrical connection

Figure 2 Schematic diagram of the tensiometer.

opening hole in the base pedestal in a standard 70 mm diame-ter oedometer, as shown in Figure 1. The soil sample was placedinside the oedometer in contact with the tensiometer, then coveredby a load cap to enable vertical loading by the piston. A neoprenemembrane was fixed to cover the soil and cap to avoid any evap-oration that could cause an increase of soil suction. The effectof the membrane is examined first. As already noted, only whenthe pore-air pressure in the soil specimen was equal to zero (i.e.the atmosphere pressure) during compression, was the measure-ment of pore-water pressure, uw, measured with the tensiometerequal to the matric suction of the soil, s. Consequently, an airpocket of 15-mm height was provided at the top of the soil sample(Figure 1).

The tensiometer used (Figure 2) is of the Imperial Collegetype (Ridley & Burland, 1993, 1996). This type of tensiometerhas been used successfully to perform suction measurementsunder laboratory conditions (Dineen & Burland, 1995; Ridley& Burland, 1996; Tarantino et al., 2000; Tarantino & Mongiovì,2001, 2002; Ridley et al., 2003). It has provided excellentperformance in terms of accuracy, measurement duration andoperating tension range (0–1.5 MPa; Tarantino & Mongiovì,2001). It was used by Cui et al. (2007) to monitor field suctionchanges. The tensiometer consists of a porous ceramic stone with

a 1.5 MPa air-entry value, a water reservoir 0.1 mm thick and astrain gauge attached to the diaphragm plate (Figure 2).

The presence of air in the water reservoir can cause cavitationof the tensiometer under suction below the maximum workingtension. This in turn makes it impossible for the tensiometerto measure suction. Therefore it is important to saturate thetensiometer well prior to use. The tensiometer was saturated in asaturation cell (80 mm in diameter and 70 mm high) as describedin Mantho (2005), using a digital pressure-volume controller. Aftereach measurement, the tensiometer was placed in the saturationcell and re-saturated at 2 MPa for 48 hours.

A calibration stage of the tensiometer was also necessaryafter the saturation stage. The voltage of the tensiometer wasrecorded while the pressure was applied in steps using a digitalpressure-volume controller, which applied a positive pressureto the tensiometer with a precision of 1 kPa. The calibrationcurve obtained in the positive range was then extrapolated in thenegative range. Tarantino & Mongiovì (2001) showed that thecalibration curve is the same in the positive and negative ranges.This calibration stage was used also to estimate the tensiometeraccuracy: it depended on the previous saturation stage and variedfrom 1 to 6.9 kPa with a mean of 2.8 kPa.

Testing programme

Soil samples were prepared by compacting soil crumbs either<2 mm or <0.4 mm, which were re-wetted to different initialwater content (wi = 0.125, 0.143, 0.16, 0.198 and 0.25 g g−1

water content) by spraying. The same mass of dry soil was usedfor all samples (103.62 ± 0.05 g). Initial compaction was carriedout directly in an oedometer to prepare the samples. The finaldimensions of the soil samples were: diameter 70 mm and height24 mm, corresponding to a bulk dry density of 1.1 Mg m−3.

After producing the soil sample, vertical stresses of 10, 20, 50,100, 200, 400 and 800 kPa were applied in a step-by-step wayby using a controlled pneumatic system. Unloading was carriedout following the same stress steps until 50 kPa and tests wereperformed under undrained conditions. The vertical displacementwas recorded using two transducers (accuracy of 0.01 mm)installed symmetrically (Figure 1). The final water contentwas determined by oven-drying at 105◦C for 24 hours. Thesemeasurements enabled the determination of bulk density, voidratio and degree of saturation of the soil samples. During loadingand unloading, suction changes were continuously monitored bythe tensiometer installed on the oedometer. Three compressiontests were performed for each initial water content condition.

Pore-size distribution measurements

Mercury intrusion porosimetry. The pore-size distribution wasstudied using the mercury intrusion technique. Mercury, a non-wetting liquid, was pushed into an air-dried soil sample underpressure (Fies, 1984; Bruand & Prost, 1987). The relationshipbetween the equivalent pore diameter (Deq, in μm) and the suction


602 K. Cui et al.

(s, in kPa) was obtained from the Jurin-Laplace law:

Deq = 4γ (cos α)/s, (1)

where γ is the interfacial tension between air and mercury(0.484 N m−1), and α is the contact angle between the soil and themercury (130◦; from Good, 1984). The pressure range was from4 to 200 000 kPa and the corresponding pore diameter was from360 to 0.006 μm. Soil volumes of approximately 3 cm3 were usedfor this study and were oven-dried at 105◦C for 24 hours prior totaking the measurements (Richard et al., 2001).

Testing programme. Soil samples for pore-size distributionmeasurements were prepared by compaction by using the sameprocedure as in the previous mechanical tests: two samples withsoil sieved at 2 mm and two samples with soil sieved at 0.4 mm.The dry bulk density of the four samples was 1.1 Mg m−3 andtheir water content was 0.16 g g−1. One sample of each sieve size(2 mm or 0.4 mm) was loaded into the oedometer in steps of 10and 20 kPa, whereas the other two were loaded in steps of 10, 20,50, 100, 200 and 400 kPa. All the samples were unloaded in onestep and then air-dried to determine pore-size distribution. Two‘crumbs’ of approximately 3 g for each sample were measuredfor this study to ensure repetition.

Results

Evaluation of the soil suction measurement procedure

It was necessary to ensure the quality of soil suction measurementsby taking precautions to avoid possible evaporation and maintainthe rate of equilibrium. The variation of suction was measured ina soil sample within the oedometer over 48 hours under a verticalstress of 200 kPa. A suction increase of approximately 1 kPa wasmeasured, showing that the anti-evaporation system using a neo-prene membrane was satisfactory. As the duration of a completecompression test was approximately 5 hours, the effect of waterevaporation on suction variation could be neglected. Figure 3presents the variation of the pore-water pressure (uw) under a ver-tical stress of 800 kPa. It was observed that once a vertical stress

-50

0

50

100

150

0 10 20 30

Time /min

uw /k

Pa

Figure 3 Pore-water pressure, uw, in relation to time under a verticalstress of 800 kPa applied at the 14th minute (soil sample with initialcrumbs <2 mm, 0.16 g g−1 water content and 1.1 Mg m−3 dry bulkdensity).

was applied, the pore-water pressure immediately increased (posi-tive value) and then decreased after approximately 5 minutes. Thisstudy focused on the subsequent equilibrium stage. As mentionedbefore, any air pressure (ua) built up during compression couldaffect uw. This problem was overcome by letting the expelledair reach the air pocket (Figure 1). Moreover, it was necessaryto ensure the equilibrium in terms of soil volume changes andwater transfer within the soil, and periods of 40 minutes for eachloading step and 5 minutes for each unloading step were used toreach equilibrium for suction and strain measurements.

Soil suction changes during loading

Figure 4(a–f) presents the variations of soil suction underloading for different initial soil suctions and different aggregatefractions (<2 mm or 0.4 mm) after equilibrium. Suction changecharacteristics in terms of the standard deviation measured fordifferent initial conditions, were as follows. Suction slightlyincreased for small vertical stresses and remained constant up to astress of 800 kPa at an initial water content of wi = 0.125 g g−1

(initial soil crumb size <2 mm, Figure 4a). Suction remainedconstant for soil samples with a water content of 0.143 g g−1

(initial soil crumb <2 mm, Figure 4b). For soil samples withwater content greater than 0.143 g g−1, suction initially remainedconstant up to a stress threshold value (σt), after which it decreased(Figure 4c–f). We determined σt by comparing the variation inthe suction value between two successive applied stresses and thestandard error of the second stress (vertical bars). If this differencewas greater than the standard error, we assumed that the first stresscorresponded to σt. The stress threshold (σt) and correspondingsuction (st) are presented in Table 1 for all the samples. It can beobserved that σt varied from 50 to 400 kPa. For the initial size ofsoil fragments, <2 mm, σt decreased with the initial water content(or increased with the initial suction), and σt for the initial size ofsoil crumbs, <0.4 mm, was smaller than that for <2 mm for aninitial water content of 0.16 g g−1 (Figure 4c and f).

We calculated the maximum equivalent diameter, Deq∗,

given by the Jurin-Laplace law (Equation (1), γ = 72.75 ×10−3 N m−1, cos α = 1), corresponding to the threshold, σt, forall initial conditions, to examine the relation between the changein soil suction with vertical stress and the change in pore sizedistribution measured by mercury porosity. The maximum equiv-alent diameter, Deq

∗, varied from 2 μm for a suction si = 14.6kPa to 22.4 μm for si = 13 kPa (Table 2).

Compression behaviour on loading

In the following, changes in void ratio and pore size under com-pression were examined in order to better understand soil suctionand threshold stress variations in relation to macroscopic soil char-acteristics such as precompression stress and degree of saturation.

Void ratio change. The variation of void ratio in relation to thelogarithm of vertical stress is illustrated in Figure 5 for one soil



(a) crumbs <2 mm, wi = 0.125 g g-1 (b) crumbs <2 mm, wi = 0.143 g g-1

120

130

140

150

160

170

180

1 10 100 1000Vertical stress / kPa

Suct

ion

/ kPa

20

30

40

50

60

70

80


Suct

ion

/kPa

(c) crumbs <2 mm, wi = 0.160 g g-1 (d) crumbs <2 mm, wi = 0.198 g g-1

0

10

20

30

40

50

60


Suct

ion

/ kPa

0

10

20

30

40

50

60


Suct

ion

/kPa

(e) crumbs <2 mm, wi = 0.250 g g-1 (f) crumbs <0.4 mm, wi = 0.160 g g-1

-10

0

10

20

30

40

50


Suct

ion

/ kPa

10

20

30

40

50

60

70


Suct

ion

/ kPa

Figure 4 Change in matric suction with vertical stressas a function of initial crumb size and water contentof soil samples. Each point corresponds to soil suctionmeasured after equilibrium of approximately 40 minuteswhilst applying a constant vertical stress. The pointsindicate the average values from three tests; the verticalbars indicate the standard errors. An initial suction at 2 kPavertical stress was considered because of the logarithmicscale.

Table 1 Results of oedometer compression tests with measurements of soil suction: values are means of three replicates ± standard deviation

Initial watercontent wi

/ g g−1

Initial soilsuctionsi / kPa

Initial size ofsoil crumbs/ mm

Final watercontent wf

/ g g−1

Stressthresholdvalue σt / kPa

Suction atthresholdstress st / kPa

Precomp-ressionpressure σp

/ kPa

Saturationratio atthresholdstress St∗

r / %

EquivalentdiameterDeq

∗ /μma

0.125 146 <2 0.125 — — 33 ± 2 — 2.00.143 66 <2 0.143 — — 24 ± 1 — 4.40.160 47 <2 0.159 400 42 ± 4 34 ± 6 66 ± 5 6.20.198 19 <2 0.197 200 22 ± 1 21 ± 2 71 ± 5 15.30.250 13 <2 0.207 50 13 ± 2 16 ± 3 73 ± 9 22.40.160 53 <0.4 0.160 100 52 ± 1 35 ± 3 43 ± 1 6.2

aCalculated with the average value of initial soil suction.


604 K. Cui et al.

Table 2 Pore volume measurements performed on soil crumbs of two sizes taken from soil samples compacted at two vertical stresses

Crumbsieve size/ mm

Appliedvertical stress/ kPa

Dry bulkdensity/ g cm−3

Void ratio e

/ cm3 cm−3Pore volumea

/ cm3 g−1

Volume ofpores Aa

/ cm3 g−1

Volume ofpores Ba

/ cm3 g−1

Volume ofpores Ca

/ cm3 g−1

Volume ofpores <Deq

∗

/ cm3 g−1

2 400 1.68 0.60 0.188 0.011 0.162 0.015 0.1490.4 400 1.65 0.64 0.207 0.006 0.182 0.019 0.1922 20 1.20 1.25 0.407 0.206 0.190 0.011 0.1270.4 20 1.18 1.28 0.401 0.060 0.323 0.017 0.127

aAverage values for two repetitions with maximum standard error of 0.001 cm3 g−1.

0.4

0.6

0.8

1.0

1.2

1.4

1.6

10 100 1000Vertical stress / kPa

e

σp

k

l

k′

Figure 5 Void ratio (e) in relation to vertical stress for a soil sample withinitial crumbs of less than 2 mm, an initial water content of 0.16 g g−1

and an initial dry bulk density of 1.1 Mg m−3. The precompression stress,σp, was determined by the intersection of two straight lines.

sample. The curve shows an over-consolidated behaviour of thesoil; that is, a slight decrease in void ratio until precompressionstress (σp), followed by a considerable decrease in void ratio.Precompression stress was determined graphically (Figure 5) byusing the standard method for geotechnical engineering (AFNOR,1997; Bardet, 1997). The first straight line with slope κ , which isassumed to be equal to slope κ ′ defined on the unloading curve, isdrawn across the initial point (σv = 10 kPa). The second straightline with slope λ is drawn across the point that has a maximumvalue of �e/�lnσv (σv = 100 kPa for these samples). Finally,the intersection of the two lines gives the precompression stress.The same calculation procedure was applied to other compressioncurves under different initial conditions. Table 1 gives all the σp

values with standard errors. It can be observed that precompres-sion stress decreased with water content, from 33 to 16 kPa, anddid not depend on initial fragment size. These measurements were

consistent with those in the literature, showing similar effects ofsoil water content on the precompression stress (Alexandrou &Earl, 1998; Horn & Fleige, 2003; Imhoff et al., 2004).

Change in saturation degree. Figure 6 shows the compressioncurve of the degree of soil saturation in relation to the logarithmof vertical stress for the same soil sample presented in Figure 6.We obtained a saturation degree value corresponding to theobserved stress threshold value (St

r), which varied from 66 to73% with the initial soil crumbs of less than 2 mm (Table 1).A value of 43% was found for the samples of 0.4-mm sieve size.In addition, we observed that the final water content, wf, wasequal to 0.207 g g−1 for initial water content, wi, of 0.25 g g−1,although the mechanical test was performed under undrainedconditions (Table 1). This difference was related to water lossfrom the sample after reaching the saturation state (Sr = 100%).This quantity was found in the edges of the oedometer cell. As thevalue of σt in this test was equal to 50 kPa and the correspondingSr value was 73%, a constant water content of 0.25 g g−1 couldbe considered below σ t.

Larson & Gupta (1980) investigated the degree of soil saturationat the transition point beyond which suction increased. They foundthat the transition occurred at the same degree of soil saturationfor the samples at different initial suctions. They established arelation between this transition degree of saturation, St∗

r , and thesoil texture based on the results from 54 soils. The relationship isas follows for CC < 33%:

St∗r = 0.364 + 0.00659CC, (2)

where CC is the clay content (%). Table 1 shows the transitiondegree of saturation, St∗

r , calculated for the different tests. Table 1

20

30

40

50

60

70

80

90


S r /

%

2030405060708090

100


S r /

%

(a) (b)

Figure 6 saturation in relation to vertical stress for the soil sampleat an initial dry bulk density of 1.1 Mg m−3 for initial watercontents of 0.16 g g−1 (a) and 0.25 g g−1 (b). The stress thresholdand the corresponding saturation rate were estimated, respectively,at σt = 400 kPa and St

r = 66% for wi = 0.16 g g−1 andσt = 50 kPa and St

r = 73% for wi = 0.25 g g−1.



shows that the threshold stress corresponds to a transition degreeof saturation, St∗

r , of approximately 70% for the samples of<2 mm and 43% for the samples of <0.4 mm.

Change in soil pore-size distribution. Samples of different sievesizes were loaded to 20 kPa and to 400 kPa. Both stressescompact the soil at, respectively, 1.2 Mg m−3 under 20 kPaand 1.65 Mg m−3 under 400 kPa in the oedometer cell. Table 2shows the measurements performed on crumbs of 3 g, whichwere sampled from the oedometer-compacted samples and oven-dried. The dry bulk densities of the crumbs were approximately1.3 Mg m−3 for samples compacted at 20 kPa and 1.75 Mg m−3

for those compacted at 400 kPa. Figure 8 shows the differences inpore volume per mass unit of the oven-dried soil as a function ofthe equivalent pore diameter. The pore volumes were calculatedfor two replicates (two crumbs taken from the same sample).The standard error was ± 0.001 of the maximum. Bruand &Prost (1987) proposed that three classes of pores (A, B andC) should be identified to analyse the curves obtained by usingmercury porosimetry. The threshold between pore-classes A andB is 40 μm, and it is 0.05 μm between pore-classes B and C.Our results (Table 2) show that we have almost the same totalvolumes of pores under 20 and 400 kPa (approximately 0.40 and0.20 cm3 g−1) for the two types of sample (<2 and <0.4 mm). At20 kPa, the majority of pores belonged to class A for the <2-mmsample and to class B for the <0.4-mm sample. Furthermore,the pores of class C changed slightly for the two sample types.When loading from 20 to 400 kPa, the pores of class A decreasedsignificantly for the two sample types and remained at this levelfor the soil <0.4 mm (0.006 cm3 g−1). At 400 kPa, the pores ofclass B decreased significantly for the <0.4-mm sample but notfor that of <2 mm.

We compared the porosity measurement of the <0.4-mmsample at a vertical stress of 400 kPa in comparison with samplesof <2 mm. Figure 7 presents the difference in pore volumedistribution between samples at 1.2 and 1.65 Mg m−3 for bothsize aggregate fractions. Figure 7 shows that there were few newsmall pores for samples of <0.4 mm at a vertical stress of 400 kPacompared with samples of <2 mm, for which there were far morenew small pores. Table 2 presents the volume of pores of meansize smaller than the D∗

eq, which is 6 μm for wi = 0.16 g g−1. Thevolume of new pores smaller that 6 μm is greater for the samplesof <0.4 mm compared with samples that were <2 mm.

Discussion

Our study focussed on the equilibrium stage for soil suctionbecause models for soil compaction usually consider stress andstrain at equilibrium. However, wheeling in cultivated soilsinvolves short loading dynamics. Different authors have inves-tigated the time response of soil suction in relation to shortloading or cyclic loading (Peng et al., 2004; Krummelbein et al.,2008). Our observations that the pore-water pressure immedi-ately increased (positive values) and then decreased (Figure 3)

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.01 0.10 1.00 10.00 100.00 1000.00Equivalent diameter / µm

Dif

fere

nce

in p

ore

volu

me

dist

ribu

tion

/cm

3 g- 1

Figure 7 Differences in pore volume distribution between samples at 1.2and 1.65 Mg m−3 of dry bulk densities as a function of pore equivalentdiameter for the soil samples of <2 mm (open squares) and <0.4 mm(black squares). Maximum standard error was 0.001 cm3 g−1.

σv = 0 Deq*

σv < σt

Figure 8 Schema of mechanisms proposed to explain suction variationunder compression at low stress (<σt). Compaction induces a decrease involume of large pores, but this decrease can be compensated for by anincrease in small pore volume so that water can be held in smaller poresunder compression, leading to a decrease in the equivalent diameter andan increase in the suction.

confirmed the results reported by Peng et al. (2004) and Tombo-lato et al. (2004), who reported measurement of soil suction in situduring compression. The dynamics of the tensiometer responsedepend on the tensiometer itself and on the soil, especially itsunsaturated hydraulic conductivity (Selker et al., 1992; Hayashiet al, 1997; Timlin & Pachepsky, 1998). In term of intensity, theimmediate pore-water pressure should equal the external appliedpressure in saturated conditions. In contrast, this transmission instress is much less in unsaturated conditions, as illustrated inFigure 3, when the transmission was 160 kPa for an applied stressof 800 kPa.

When modelling soil compaction caused by the passage of agri-cultural vehicles, it has been generally assumed that total stresscan be used and that matric suction remains constant (or watercontent remains constant) under the undrained conditions gen-erally assumed with a short loading time (<0.1 s). Our resultsshow that for a static compression under undrained conditions,there is a domain of stress where this assumption holds. Thisdomain is delimited by a threshold stress (σt), which depends onthe initial soil suction: the greater the initial suction, the greater


606 K. Cui et al.

the stress threshold. These results are in agreement with thoseobtained by Larson & Gupta (1980), Wulfsohn et al. (1998) andda Veiga et al. (2007) but differ from those obtained by Tarantino& Tombolato (2005). Larson & Gupta (1980) measured the changein matric suction during uniaxial compressions of 54 soils at differ-ent initial suctions ranging from 5 to 60 kPa. They observed thatunder loading the matric suction increased to a certain value andthen decreased. The maximum increase in soil suction (20 kPa)was obtained under small initial suctions of 20 and 40 kPa. Ourresults show a similar characteristic for an initial soil suctionof 146 kPa, with a maximum increase of approximately 10 kPa.Wulfsohn et al. (1998) studied the influence of matric suction onsoil strength by performing triaxial tests on a sandy clay loam at aninitial bulk density of 1.2 Mg m−3 and an initial suction of 50 kPa.The applied confining pressure varied from 1.5 to 250 kPa. Theyobserved that when the confining pressure was less than 50 kPa,the variation in matric suction was less than 8 kPa. The suctiondecreased significantly in the case of larger confining pressures.Da Veiga et al. (2007) investigated the effect of long-term tillage(no-tillage, chisel-ploughed and conventional tillage treatments)on the soil suction change under compression for clay soils (clayfraction up to 700 g kg−1). The matric suction first increased andthen decreased for the three treatments but this first increase wasless pronounced for the no-tillage treatment. Tarantino & Tombo-lato (2005) reported measurements on a clay soil (clay fraction800 g kg−1) at different initial bulk densities ranging from 1.05 to1.3 Mg m−3, for which soil suction increased systematically dur-ing compression. It appears therefore from different studies that atransition point below which the assumption of constant suctionholds can be observed. Questions remain about the stress thresh-old value, σt, with respect to the vertical stress usually applied byagricultural machinery (<400 kPa) and the factors affecting σt.

We investigated the relationship between the soil saturation andthe transition point beyond which suction increased according toLarson & Gupta (1980), who found that the transition occurredat the same degree of soil saturation for samples initially atdifferent suctions. By applying Larson & Gupta’s relationshipto our soil (Equation (2)), we obtained a more or less constanttransition degree of saturation (St∗

r ) of 70% for samples initiallyat different suctions, for the samples of <2 mm (Table 1). Thisconfirms that the degree of saturation is a factor that affects thethreshold stress. However, the transition degree of saturation (St∗

r )was noticeably different for the <0.4-mm sample (St∗

r = 43%),so soil structure should also be taken into account when predictingthreshold stress. Furthermore, the results reported by Tarantino& Tombolato (2005) on a clay soil (clay fraction 800 g kg−1)

suggest that the relation proposed by Larson & Gupta (1980) maybe not valid for soils with very large clay content.

Peng et al. (2004) proposed that soil suction variation duringcompression should be correlated with precompression stress (σp),as this depends on soil texture, soil structure and soil suction.Table 1 shows that for w = 0.16, 0.198 and 0.25 g g−1, thresholdstress (σt) was, respectively, 400, 200 and 50 kPa, much largerthan the corresponding precompression stress (34, 21 and 16 kPa).

This shows that the threshold stress, σt, would be smaller withsmall precompression stresses, but the domain of constant soil suc-tion was larger than that delimited by the precompression stress.

The interpretation of the changes in soil structure accompaniedby changes in soil suction during compression can be madequalitatively at the pore scale. Intuitively, the compaction atconstant water content can be thought of as follows: a globaldecrease in pore space should cause a decrease in soil suction;that is, the soil water volume remains constant while pore sizedecreases and pores disappear under compression. However, thisview does not consider the fact that the change in pore size is nothomogeneous for all pore sizes under compression.

Our results obtained by mercury porosimetry showed thatmechanical compression induced a decrease in large pore vol-ume and an increase in small pore volume with a limit betweendisappearance and creation for pore size of a few microns. Thepore-size distribution did not change in a uniform fashion. Ourobservations are consistent with observations showing that com-paction may decrease large pores (>10 μm) and increase smallpores (0.1–10 μm) in soils of varying texture in the range ofmechanical stresses applied in agriculture (<400 kPa) (Bruand &Cousin, 1995; Richard et al., 2001; Tarawally et al., 2004; Kutíleket al., 2006). Our hypothesis is that the domain of constant suctionresults from competition between the redistribution of soil waterin small pores created by compression that tends to increase soilsuction and a global decrease in pore space that tends to decreasesoil suction. This assumption can be examined by comparing qual-itatively the change in soil suction with vertical stress observedfor samples at an initial suction (si) of 47 kPa for both soil frac-tions considered here (<2 and <0.4 mm) (Figure 5c and f) inrelation to the change in pore-size distribution measured by mer-cury porosimetry for compacted samples obtained after verticalstresses of 20 and 400 kPa. There were fewer new small poresin samples of <0.4 mm at a vertical stress of 400 kPa comparedwith samples of <2 mm. This effect could explain the differencein the values of stress threshold (σt = 400 kPa for <2-mm and100 kPa for <0.4-mm samples at an initial suction of 47 kPa).At an initial suction of 47 kPa, one can assume that water isessentially located in pores with a maximum equivalent diame-ter, Deq

∗, of approximately 6 μm. This pore size falls within therange of pore size at which creation of new pores was observed.Thus, soil suction is assumed to increase or decrease dependingon the changes in maximum equivalent diameter, Deq

∗, with themechanical stress level, as shown schematically in Figure 8. At avertical stress of 400 kPa, suction remains constant for the <2-mmsample because there are still enough new pores that could con-tain the water originating from the macropores that are lost, sothat the maximum equivalent diameter, Deq

∗, remains constant.In contrast, suction decreases for the <0.4-mm sample becausemechanical loading did not continue to create enough new pores.The decrease of stress threshold, σt, with decreasing initial soilsuction could also be interpreted in terms of water redistributionat the pore scale. A soil suction of 13 kPa corresponds to anequivalent pore diameter of approximately 22 μm (Table 1). At



this suction, the creation of more small pores by loading could beexpected (not measured), but their volume would not be sufficientto contain the soil water from the destroyed pores, and thus soilsuction decreases (Figure 4e). In contrast, at a large soil suction of146 kPa, new small pores are assumed to lead to a redistributionof water into smaller pores that tend to increase soil suction. Thisagrees with the increase in soil suction at small vertical stressesobserved for an initial large soil suction (Figure 4a).

This analysis of the changes in soil suction during compressionin relation to change at the pore scale is restricted to qualitativeconsiderations. Indeed, quantitative analysis is limited by meth-ods used for sampling and drying in this study. As the volumeof ‘crumbs’ used in mercury porosimetry was small (mm3) incomparison to the volume of soil in the oedometer cell (cm3),sampling for mercury porosimetry is expected to represent, at leastpartially, the soil structure of oedometer samples. However, thecrack and macropore structure may be under-estimated by mer-cury porosimetry. Secondly, the technique used to dry the soilbefore porosimetry induces unavoidable shrinkage, which can belimited by freeze-drying (Delage et al., 1996). Both sampling anddrying can explain discrepancies in dry bulk density of the crumbsin comparison with the initial compacted volume, as reported inTable 2. Further investigation into the relationship between macro-scopic threshold, σt, and pore size distribution in relation to soilsuction involves the difficult problem of quantifying the change inpore space caused by compaction in relation to retention properties(Pagliai et al., 2003; Hajnos et al., 2006; Schaffer et al., 2007).

Conclusion

Variations of soil suction under static compression were inves-tigated by using an oedometer with soil suction measurements.For initial suction greater than 20 kPa, matric suction remainedquasi-constant under a stress threshold (σt), which increased withincreasing initial soil suction and with increasing sieve size. Forinitial suction greater than 20 kPa, the values of σtfell within therange of mechanical vertical stress generally exerted by agricul-tural machines (<400 kPa). This corroborates the assumption ofconstant suction during deformation, which is usually adopted inmodelling soil compaction caused by machinery traffic but ques-tions its validity for soils that are close to saturation.

The variations in suction during loading were qualitativelyanalysed at the pore scale by measuring their distribution bymercury porosimetry. Loading deformed the pores; large poresdecreased in size and created new small pores. If the degree ofwater saturation is low (as in the case of low water content and alarge suction), the volume of the smaller pores created is sufficientto balance the water flow caused by the destruction of other pores,and thus soil suction tends to increase or remain constant. Thisis the case for σ < σt. For larger initial degrees of saturation (asin the case of larger water contents or greater vertical stress), thenew small pores no longer compensate for the loss of larger poresand any further loading leads to a decrease in suction. This isthe case for σ > σt. Both the aggregate size, as shown in this

study, and the clay content, as shown by Larson & Gupta (1980),change the stress threshold, probably because of different changesin pore-size distribution as the soil deforms. Further measurementson aggregate fractions larger than 2 mm should be performed toprovide a greater knowledge of this effect of structure on thestress threshold, which is an important component required formodelling effects in soils.

Acknowledgements

We thank F. Bornet (UR1158, INRA, France) and E. Delaure(CERMES, ENPC, France) for their assistance in the developmentof the mechanical set-up, and I. Cousin and O. Josiere (UR0272Science du Sol, Orleans, INRA, France) for pore-size distribu-tion measurements. The authors thank the French GESSOL2 Pro-gramme of the Ministry of the Environment and the ADD Programof the National Research Agency for providing their support tothe DST project (Soil degradation due to compaction), of whichthis work is a part.

References

AFNOR 1997. Sols: Reconnaissance et Essais. Essai Oedometrique. Partie

1: Essai de Compressibilite sur Materiaux Fins Quasi Satures Avec

Chargement Par Paliers. XP P 94-0909-1, Paris, France.Alexandrou, A. & Earl, R. 1998. The relationship among the pre-

compaction stress, volumetric water content and initial dry bulk densityof soil. Journal of Agricultural Engineering Research, 71, 75–80.

Alonso, E.E., Gens, A. & Josa, A. 1990. A constitutive model for partiallysaturated soils. Geotechnique, 40, 405–430.

Bardet, J.P. 1997. Experimental Soil Mechanics. Prentice-Hall, Inc., UpperSaddle River, New Jersey, USA.

Bruand, A. & Cousin, I. 1995. Variation of textural porosity of a clay-loamsoil during compaction. European Journal of Soil Science, 46, 377–385.

Bruand, A. & Prost, R. 1987. Effect of water content on the fabric of asoil material: an experimental approach. Journal of Soil Science, 38,461–472.

Burland, J.B. 1965. Some Aspects of the Mechanical Behaviour of PartlySaturated Soils, Proc. Conf. on Moisture Equilibria and MoistureChanges in Soil Beneath Covered Areas (ed. G.D. Aitchison), pp.270–278. Buttersworth, London.

Cui, Y.J. & Delage, P. 1996. Yielding and plastic behaviour of anunsaturated compacted silt. Geotechnique, 46, 291–311.

Cui, Y.J., Tang, A.M., Mantho, A. & De Laure, E. 2007. Monitoringfield soil suction using miniature tensiometer. Journal of Geotechnical

Testing, 31, 95–100.Delage, P., Audiguier, M., Cui, Y.J. & Howat, M.D. 1996. Microstructure

of a compacted silt. Canadian Geotechnical Journal, 33, 150–158.Dineen, K. & Burland, J.B. 1995. A new approach to osmotically

controlled oedometer testing. In: Unsaturated Soils, Proceedings of the

1st International Conference on Unsaturated Soils (eds E.E. Alonso &P. Delage), pp. 459–465. Balkema, Rotterdam.

Fies, J.C. 1984. Analyse de la repartition du volume des pores dans lesassemblages argile squelette: comparaison entre un modele d’espaceporal textural et les donnees fournies par la porosimetrie au mercure.Agronomie, 4, 319–355.


608 K. Cui et al.

Good, R.J. 1984. The contact angle of mercury on the internal surfaces ofporous bodies. A foot note to D.N. Winslow’s review of porosimetry.Surface Colloid Science, 13, 283–287.

Gysi, M. 2001. Compaction of a Eutric Cambisol under heavy wheeltraffic in Switzerland: field data and a critical state soil mechanics modelapproach. Soil & Tillage Research, 61, 133–142.

Hajnos, M., Lipiec, J., Swieboda, R., Sokołowska, Z. & Witkowska-Walczak, B. 2006. Complete characterization of pore size distribution oftilled and orchard soil using water retention curve, mercury porosimetry,nitrogen adsorption, and water desorption methods. Geoderma, 135,307–314.

Hayashi, M., van der Kamp, G. & Rudolph, D.L. 1997. Use of tensiometerresponse time to determine the hydraulic conductivity of unsaturatedsoil. Soil Science, 162, 566–575.

Horn, R. & Fleige, H. 2003. A method for assessing the impact of load onmechanical stability and on physical properties of soils. Soil & Tillage

Research, 73, 89–99.Imhoff, S., Da Silva, A.P. & Fallow, D. 2004. Susceptibility to com-

paction, load support capacity, and soil compressibility of Hapludox.Soil Science Society of America Journal, 68, 17–24.

Jennings, J.E. & Burland, J.B. 1962. Limitations to the use of effectivestresses in partly saturated soils. Geotechnique, 2, 125–144.

Kirby, J.M. 1994. Simulating soil deformation using a critical-state model.I. Laboratory tests. European Journal of Soil Science, 45, 239–248.

Krummelbein, J., Peth, S. & Horn, R. 2008. Determination of pre-compression stress of a variously grazed steppe soil under static andcyclic loading. Soil & Tillage Research, 99, 139–148.

Kutílek, M., Jendele, L. & Panayiotopoulos, K.P. 2006. The influence ofuniaxial compression upon pore size distribution in bi-modal soils. Soil& Tillage Research, 86, 27–37.

Larson, W.E. & Gupta, S.C. 1980. Estimating critical stress in unsaturatedsoils from changes in pore water pressure during confined compression.Soil Science Society of America Journal, 44, 1127–1132.

Mantho, A.T. 2005. Echanges sol-atmosphere application a la secheresse

(France). PhD thesis, Ecole Nationale des Ponts et Chaussees, Paris.O’Sullivan, M.F. & Simota, C. 1995. Modelling the environmental

impacts of soil compaction: a review. Soil & Tillage Research, 35,69–84.

Pagliai, M., Marsili, A., Servadio, P., Vignozzi, N. & Pellegrini, S. 2003.Changes in some physical properties of a clay soil in Central Italyfollowing the passage of rubber tracked and wheeled tractors of mediumpower. Soil & Tillage Research, 73, 119–129.

Peng, X.H., Horn, R., Zhang, B. & Zhao, Q.G. 2004. Mechanisms of soilvulnerability to compaction of homogenized and recompacted Ultisols.Soil & Tillage Research, 76, 125–137.

PLAXIS 1998. Finite Element Code for Soil and Rock Analysis, Version7. Balkema, Rotterdam.

Richard, G., Cousin, I., Sillon, J.F., Bruand, A. & Guerif, J. 2001. Effectof compaction on the porosity of a silty soil: influence on unsaturatedhydraulic properties. European Journal of Soil Science, 52, 49–58.

Richards, B.G. 1992. Modelling interactive load-deformation and flowprocesses in soils, including unsaturated and swelling soils. In

Proceedings of the 6th Australia-NZ Conference on Geomechanics ,Christchurch, pp. 18–37.

Ridley, A.M. & Burland, J.B. 1993. A new instrument for the measure-ment of soil moisture suction. Geotechnique, 43, 321–324.

Ridley, A.M. & Burland, J.B. 1996. A pore pressure probe for the in situmeasurement of a wide range of soil suctions. In: Advances in SiteInvestigation Practice (ed. C. Craig), pp. 510–520. Thomas Telford,London.

Ridley, A.M., Dineen, K., Burland, J.B. & Vaughan, P.R. 2003. Soilmatrix suction: some examples of its measurement and application ingeotechnical engineering. Geotechnique, 53, 241–253.

Schaffer, B., Stauber, M., Muller, R. & Schulin, R. 2007. Changes in themacro-pore structure of restored soil caused by compaction beneathheavy agricultural machinery: a morphometric study. European Journal

of Soil Science, 58, 1062–1073.Selker, J., Leclerq, P., Parlange, J.-Y. & Steenhuis, T. 1992. Fingered flow

in two dimensions: 1. Measurement of matric potential. Water ResourcesResearch, 28, 2513–2521.

Tarantino, A. & Mongiovì, L. 2001. Experimental procedures andcavitation mechanisms in tensiometer measurements. Geotechnical &Geological Engineering, 19, 189–210.

Tarantino, A. & Mongiovì, L. 2002. Design and construction of a ten-siometer for direct measurement of matric suction. In: UnsaturatedSoils, Proceedings of the 3rd International Conference on Unsatu-

rated Soils (eds J.F.T. Juca, T.M.P. de Campos & F.A.M. Marinho),pp. 319–324. Balkema, Recife, Brasil.

Tarantino, A. & Tombolato, S. 2005. Coupling of hydraulic and mechan-ical behaviour in unsaturated compacted clay. Geotechnique, 55,307–317.

Tarantino, A., Mongiovì, L. & Bosco, G. 2000. An experimental investi-gation on the independent isotropic stress variables for unsaturated soils.Geotechnique, 50, 275–282.

Tarawally, M.A., Medina, H., Frometa, M.E. & Alberto Itza, C. 2004.Field compaction at different soil-water status: effects on pore sizedistribution and soil water characteristics of a Rhodic Ferralsol inWestern Cuba. Soil & Tillage Research, 76, 95–103.

Timlin, D. & Pachepsky, Y. 1998. Measurement of unsaturated soilhydraulic conductivities using a ceramic cup tensiometer. Soil Science,163, 625–635.

Tombolato, S., Tarantino, A. & Mongiovì, L. 2004. Suction induced bystatic compaction, Proc. Int. Conf. From Experimental Evidence towardsNumerical Modeling of Unsaturated Soils, Weimar, Germany, 2003 (ed.T. Schanz), pp. 101–110. Springer, Berlin.

da Veiga, M., Horn, R., Reinert, D.J. & Reichert, J.M. 2007. Soilcompressibility and penetrability of an Oxisol from southern Brazil,as affected by long-term tillage systems. Soil & Tillage Research, 92,104–113.

Wheeler, S.J. & Sivakumar, V. 1995. An elasto-plastic critical frameworkfor unsaturated soil. Geotechnique, 45, 35–53.

Wulfsohn, D., Adams, B.A. & Fredlund, D.G. 1998. Triaxial testingof unsaturated agricultural soils. Journal of Agricultural Engineering

Research, 69, 317–330.


Download - Soil compaction by wheeling: changes in soil suction ... · Soil compaction by wheeling: changes in soil suction caused by compression ... oedometrique standard´ ´equip ´ed’untensiom

Top Related