TRANSPORT RESEARCH LABORATORY

E[®E~Eo ~hJ§

TITLE

by

Soil compaction at low moisture contentsin Kenya

M J O'Connell, et al

Overseas CentreTransport Research LaboratoryCrowthomne Berkshire United Kingdom

C

-

O'CONNELL, M J et al, 1987. Soil compaction at low moisture contents

in dry areas in Kenya. In: AKINMUSURU, J 0 et al (Eds). Soil

Mechanics and Foundation Engineering. Ninth Regional Conference for

Africa, Lagos, September 1987, Volume 1. Rotterdam: A A Balkema,211-226.

9th Regional Conference for Africa on Soil Mechanics and FoundationEngineering/Lagos/Septemfber 1987

SOIL COMPACTION AT LOW MOISTURE CONTENTS IN DRY AREAS IN KENYA

M J O'CONNELL*, 3 H G WAMBURA** AND D NEWILL*

* Transport and Road Research Laboratory, United Kingdom

** Ministry of Transport and Communications, Ke nya

ABSTRACT: A joint research project between the Ministry of Transport andCommunications, K(enya and the Transport and Road Research Laboratory. UK is a study ofsoil compaction at low moisture contents in arid areas. This paper describes part ofthe study in which an investigation has been carried out on a well-graded clayey graveldesignated as a road base material for a road project in NW Kenya. The investigationcomprised laboratory testing, pilot and full-scale trials to examine the density andstrength characteristics of the material at low moisture contents. The results werecompared with those obtained when compaction was carried out under normal requirementsat optimum moisture content.

It was shown that high densities were achieved at low moisture contents both inlaboratory tests and in the field trials and that vibrating compaction methods werepreferred. Conventional construction plant and operating techniques were successfullyapplied and specifications for compaction in Kenya were readily exceeded. Field CBRmeasurements for the base constructed at low moisture contents were lower than thoseobtained in the laboratory. Analysis of pavement strength parameters incl~udingdeflection, radius of curvature and structural number showed that the trial sectionswere different but indicate that the designs were adequate for a life of 0.5 millionequivalent standard axles. The road trials constructed with bases at differentinitial moisture contents have performed equally well for two years but furtherinformation on the longer term performancecan be made on the use of dry compaction.

1. INTRODUCTION

Soil compaction at constant compactiveeffort and different moisture contentsgenerally produces a dry density-moisturecontent relation with a well-definedmaximum dry density and optimum moisturecontent similar to that shown in Fig 1.At very low moisture contents and atconstant compactive effort the density ofmany soil types will increase in a wayshown in Fig 2. For most cases the natur-al moisture contents of soils are closeenough to the optimum for only relativelysmall adjustments to be required in orderfor maximum density 'to be achieved. Inhot dry climates, however, this is not soand problems often occur if largequantitites of water have to be added inareas where it is both scarce and expens-ive. The fact that high densities can beachieved at low moisture contents andthat this has only rarely been exploited,

is required before final recommendations

has led to a joint research project beingset up between the Transport and RoadResearch Laboratory, UK, and the Ministryof Transport and Communications, Kenya.It was appropriate to undertake the workin Ken~ya as approximately fifty five percent of the country has an annual rainfallof between 500 and 250mm a year and fora further twenty per cent the rainfallis less than 250mm. Also in these areasa number of major road projects wereplanned where water for construction pur-poses was known to be a severe problem.

The research project has involvedstudies of fine-gra~ined soils typical ofthose used in road embankments and sub-grades, and of coarse-grained gravellysoils used in sub-bases and bases. Thispaper describes an investigation of onematerial, a well-graded clay gravel,designated as base material for the roadthat was being upgraded from gravel to

Soil compaction at low moisture contents in dry areas in KenyaM .J O'Cnnnefl , J H C Wimhujra ;rnd D Nc-wi l]

1

1

Maximumdry/densitytl--/

Constantcompactiveeffort

Moisture content

Fig. 1 Conventional compaction curve

Moisture content

Fig. 2 Extended compaction curve for lowmoisture contents

Fig. 3 Trial location and rainfall map of Kenya

Soil compaction at low moisture contents in dry areas in KenyaM J O'Connell, J H G Wambura and D Newill

*0

A

0

BD

bitumen surfaced standard from MarichPass to Lodwar in north-western Kenya.The 200 km route passed through the aridor semi-arid region of the Rift Valley(see Fig 3). The study of the materialwas carried out in three stages. Thesewere: -

i) Laboratory tests to determine drydensity-moisture content relationsobtained by different compaction methodsto see if high densities could be obtain-ed at low moisture contents.ii) Pilot-scale compaction trials todetermine the most effective type ofroller for the material especially at lowmoisture contents and to see if laboratorycompaction results predicted fieldbehaviour.iii) Full-scale road trials to comparethe performance of the road base materialcompacted at a range of moisture contentsunder the influence of traffic andclimatic conditions during the life ofthe road.

The performance of the full-scale trialshas been monitored so far for two yearsand results are presented of surface andsub-surface measurements made at andsince construction. These have comprisedvisual inspections, surface deformation,pavement deflection including radius ofcurvature, pavement strength and moistureconditions of the pavement layers. Someof the factors associated with compactionat low moisture contents are alsodescribed.

2. MATERIALS USED IN THE STUDY

2.1 Description and classification

The materials used in the study werealluvial quartzitic clayey gravels assign-ed for use in the road base of the 200 kmlong Marich Pass to Lodwar road. Accord-ing to the British Soils ClassificationSystem they are described as well-gradedclayey gravels with the symbols GWC (BSI1981). The material used for the initiallaboratory tests and for the pilot-scalecompaction trials was from the southernend of the road project, from a borrow pitat km 25 from Marich Pass. That used forthe full-scale- road trials was from a,borrow pit at km 19 from Lodwar at thenorthern end of the project.

The gradings and plasticities of thematerials from the different sources areshown in Fig 4 and indicate that thematerials have a similar classification.

They were generally the same as the mat-erials used for the road base throughoutthe project.

2.2 Density-moisture content relationsfrom laboratory tests

initial dry density-moisture content re-lations were determined using the materialfrom km 25 from Marich Pass. Tests werecarried out according to British Standardmethods (BSI 1975) using three differentlevels of compaction, the 2.5 kg and 4.5kgrammer dynamic compaction methods and thevibrating hammer method. The range ofmoisture contents extended from the verydry natural moisture content of the soilof between 1 and 2 per cent to the sat-urated condition of about 11 per cent.The results shown in Fig 5 clearlyindicated the potential for achievinghigh dry densities at low moisturecontents. For the vibrating hammer and4.5 kg rammer methods the results wereat least equal to 95 per cent of themaximum dry density achieved in the 4.5kgrammer method at the optimum moisturecontent, which is the specification forfield compaction in Kenya. It was becauseof this that a decision was made toproceed with the pilot-scale compactiontrials. Further laboratory *tests werecarried out on samples from the otherborrow pit at km 19 from Lodwar that wasto provide material for the full-scaleroad trials. The results of these testsare discussed later in the paper.

3. PILOT-SCALE COMPACTION TRIALS

The purpose of the pilot-scale trials was;to assess the performance of differentcompaction plant, to compare the drydensity-moisture content relationsobtained with the results of the lab-oratory tests, and to determine whetherfull-scale trials should be conducted.The types of plant, designated ascompactors 1 to 4, were two vibrating andtwo pneumatic tyred rollers and these arespecified in accordance with normalpractice either by mass per metre width ormass per wheel in Table 1. (Departmentof Transport 1986). The clayey gravel,material was prepared at a range ofmoisture contents from between 1 and 2per cent to a maximum of 8.5 per cent.At each moisture content a section 150mmthick was laid and subdivided into fourcompaction bays. Each bay was then com-pacted with six passes of one of therollers. After compaction the dry density

Soil compaction at low moisture contents in dry areas in KenyaM J O'Connell, 3 H G Wambura and D Newill

3

10C

Particle size mmm o 0 In0 C nI

( 0 - ? 1 0'~ 0 CO~ '0

FINE MED CO MS FInEo MED COA.

1. Pilot scale compaction trial 2. Full scale road trialI ---Grading envelope (KDMV)

Fig. 4 Road base :particle size distribution and limitswas determined by the sand replacementmethod (BSI 1975). The best overallperformance throughout the moisture rangewas obtained by compactor No 1, thevibrating roller, which had the greatestmass per metre width. The dry density-moisture content relation obtained forthis roller is shown in Fig 6 where it iscompared with the laboratory compactioncurves. It can be seen that the normallyspecified density for road bases isachieved at both the lowest moisturecontent and the optimum moisture content.

Table 1. Compactors used intrials

pilot-scale

Compactor Static mass! Mass!type metre width wheel

Mgs Mgs

1.Vibratory 3.42.Vibratory 2.43.Pneumatic - 1.34.Pneumatic - 3.5

4. FULL-SCALE ROAD *TRIALS

The aim of the trial was to examine theperformance of the road base, compactedat different moisture contents, under thetraffic and climatic conditions of theregion. A site for the trials waslocated at Km six south of Lodwar (seeFig 3) on the Marich Pass to Lodwar road.

At Lodwar the average annual rainfall isapproximately 200 mm and the average dailymaximum and minimum temperatures are 35 0and 24 0C. The water table is not near thesurface.

4.1 Design and layout of the trial

The results of the laboratory tests andthe pilot-scale trials provided the basisfor planning and designing the full-scaletrials. The recommendations of the KenyaDesign Manual (MOTC 1981) were generallyfollowed for pavement design and weresimilar to that being used for the con-struction of the upgraded Marich Pass toLodwar road. The subgrade material was amedium-fine non-plastic silty sand, BSCSgroup symbol SM, which had a design CBRstrength in excess of 30 per cent at thespecified density of 100 per cent of thedensity obtained in the 2.5 kg rammermethod of compaction. The Kenya DesignManual subgrade classification was S6.The estimated traffic was within the DesignManual category T5 of 0.5 million equiva-lent standard axles during a 15 yeardesign life so that pavement structuretype 1 was appropriate. This permitted asub-base of natural material and a naturalgravel base with a double surface dressing.The strength of the subgrade meant that asub-base was not required.

The clayey gravel used for road baseconstruction and selected for the trialdid not satisfy the Kenya Design Manualspecification of soaked CBR of 80 per centat the specified density. However in dryareas with good drainage the Manual doespermit a lower specification for road basematerials if they have less than 35 percent passing the 75 micron sieve, aplasticity index of less than 20 per centand a soaked CBR of greater than 50 percent. The clayey gravel met these re-quirements.

The road trial was designed to havethree sections each 100 m long with theirroad bases constructed at differentmoisture contents. These were the naturalmoisture content of 1.7 per cent, the,optimum.mois~ture content of 6.7 per centand an intermediate moisture content of3.5 per cent which was the moisture con-tent of the lowest density achieved in thelaboratory and pilot-scale trial. In allother aspects the design of the sectionswas the same. The section at optimummoisture content, which represented normalconstruction practice, was the controlagainst which the other two sections were

Si1 r~p i --n ;il 1 --w mni (lsr -nntprit., i n dry )ir~nis i n KrnT-vi

2.15

2.1 0 1-

2.05 F-

2.00 H-

1.95 h-

1.90 h-

.85 1-

1.80

0 2 4 6 8 1 0 1 2 14 16

Moisture content (per cent)

Fig. 5 Dry density/moisture content relationshipfor road base gravel (pilot scale trials)

to be compared. The section compacted atthe intermediate moisture contentprovided the opportunity to asses thebehaviour of soils whose natural moisturecontents are such that it may bedifficult to achieve the density spec-ified for construction.

In the paper the section at optimum isdescribed as Section A, the section atthe intermediate moisture content asSection B and the dry section as SectionC.

The lay-out of the experiment and thecross-section design is shown in Fig 7.

4.2 Construction of the full-scale trial

4.2.1 The subgrade

In the preparation of the sand subgradethe residual gravel from the old gravelsurfaced road was removed. During re-shaping of the subgrade which was at anatural moisture content of two per centsome shearing occurred. Further com-paction in this dry condition wascarried out with a vibrating rolleroperating in the vibrating and non-vibrating mode. Although specifieddensities were achieved it was clearthat the sheared area remained looserand weaker than adjacent unsheared areas.It was decided to rely on the confinementof the overlying base layer to ensure

0:

~E

EE

a,

U~

0.

E0U

W

.Z

01

c)

CL

99

97

95

93

91

89

87

85

0 2 4 6 8 1 0 1 2 14

Moisture content (per cent)

Fig.6 Field dry density/moisture content relationshiprelative to laboratory methodsjfor compactor 1.

that the specified design of CBR greaterthan 30 per cent would be obtained.This was confirmed during subsequentfield testing (see Fig 8). The particle-size distribution of the silty sand sub-grade material is shown in Fig 9.

4.2.2 The road base and surfacing

The clayey gravel base material washauled to the site and dumped by tippertrucks. Spreading was carried out bymotor grader and in the case of thesections at the intermediate and optimummoisture content water was added from awater tanker fitted with a spray bar.Water losses by evaporation during themixing process were 30 per cent of thetotal volume added so that, for thesection prepared at optimum moisturecontent. 30,000 litres were required.Half of this amount was added to thesection at the intermediate moisturecontent.

The road base was compacted in asingle layer to give a minimum thicknessof 150 mm. Nine passes of a vibratingroller of static mass 2.3 Mg/in widthwere applied based on the results of thepilot scale trials and recommendationsgiven in the UK Specifications forHighway Works (Department of Transport1986).

The sections were sealed with a double

,a

0

1

C:" i 1 i ,,, -, � 1 ---- --- f ------ --- ---I --- - --- � .. . .S

Longitudinalsection

Construction

moisture content

Section code

Surfacing :double surface dressing

Roadbase

Intermediate: 3.4% Optimum: 6.7% Dry: 1.7%

B A c

Subgrade

Non plastic silty sand

Cross section

5.5 m running width

.46.4m sealed width-

Sand subgrade

Fig. 7 Layout of full scale trials

,cl

Em-

CB R per cent

Fig. 8 Distribution of subgrade CBR's (DCP method)

surface dress~ing using a bituminousbinder MC 3000 sprayed at a rate of 1 .71/rn2 for the first layer and 1. 111/nt forthe second layer. The higher applicationrate of the first seal was because noprime coat was applied to the base beforesurface dressing. A continuously gradedbasalt aggregate was used as chippingsfor both seals each with the same maximumsize of 18 mm. The seal was effectivelyan 'Ottadekkel type which has been usedon a number of lightly trafficked roadsin Kenya (Thurmann-Moe and Ruistuen 1983).

4.3 Construction control

4.3.1 Density and moisture contentmeasurements

After compaction of the base, for eachsection, eleven dry density determinationsusing the sand replacement method andfive moisture content measurements weremade. The results are shown in Table 2.The distribution of density resultsrelative to the maximum achieved in the4.5 kg rammer method is shown in Fig 10.The mean density and moisture contentresults for each section are shown in

6

'150mm

0

200

E 400E-f 600CL

0 800

1000

1200

100Particle size)Jmn

900 0o r 0Ln C 0::0

mO -1 c co C~ m N

Fig. 9 Particle size distribution ofsubgrade material

80

CW

CL

U~

ZE

0.

70

60

50

40

30

20

1 0

093 94 95 96 97 98 99 100 101

Relative density per cent (4.5kg rammer method)

Fig.10 Distribution of field densities at construction

densities were achieved for Section C withmean values of 100 per cent relative com-paction whereas the other two sectionshad mean values just exceeding the minimumspecified value.

4.3.2 In-situ bearing strength measure-ments

Table 2. Dry densities and moisturecontents of full-scale trials

Dry Density Moisture Content

Section Mg/rn3 per cent

Mean SD Mean SD

A 2.054 0.03 6.7 0.4B 2.045 0.03 3.6 0.5c 2.125 0.03 1.7 0.2

Fig 11 together with the.CBR and drydensity/moisture content relations fromfour laboratory compaction tests. Thedensity/moisture content relations wereprepared from the material used for thefull scale trials and the series of iso-CBR lines were derived from unsoakedCBR tests carried out on the samples com-pacted in the laboratory.

The results show that the highest field

Although the highest density was obtainedin the dry compacted Section C it wasimportant to obtain information about thestrength of the different bases. A rapidtest with the Clegg Impact Hammer (Clegg1976) was used which gave values thatcould be correlated with CBR. Tests werecarried out at intervals during compactionafter 3,5 and 7 passes and on completionof compaction at 9 passes. The resultsare shown in terms of CBR in Table 3where each value is the median of 25 testr..The distribution of the results of testsmade on completion of compaction is shownin Fig 12. Table 3 shows that thestrength increased with the number ofpasses of the roller but was considerablylower for the dry compacted section.

4.3.3 Base thickness measurements

The thickness of the road base was deter-mined from optical level measurementstaken before and after the base materialwas placed and compacted. More than 55

7

100

CCC

cl

C7)

CL

1

Field dry density/moisture

2.20 r G) content results

- - - Lines of equal CBR value(ISO CBR)

Fig. 11 Laboratory CBR/dry density/moisture content relationship for roadbase gravel (full scale trial) with field results

\a-

2.15

2.10

C;7

2.0

2.00

1.95

1 .90

:a0

E,-

EE

a,

0Ut

0~

1 2 3 4 5 6 7 8 9 10 1 1Moisture content (per cent)

Table 3. Relation between increasingcompaction and CBR

Compaction Median CBR value %No. of (Clegg Impact Hammer)passes _ _ _ _ _ _ _ _ _ _ _ _ _ _

SECTION

A B c

3 40 47 235 42 50 287 - 52 309 78 66 44

0)a,

cim

100

90

80

70

60

50

40

30

20

10

120 140 160 180 200 220 240Thickness (mm)

Fig. 13 Distribution of roadbase thickness

100

90 I-

80 -

70 F-

d.

0

mEU

60 ~.

so -

40 I-

30 I-

20 I-

10 I

00 20 40 60 80

CB R per cent100 120

Fig. 12 Distribution of roadbase strength atconstruction (Clegg Impact Hammer)

values were obtained for each section andthe distribution of road base thicknessmeasurements is shown in Fig 13. Themedian values for the sections were be-tween 177 and 187 mm and more than 90 percent of values were equal to or greaterthan the minimum 150 mm thickness requiredin the design.

5. MONITORING PROGRAMME

The monitoring programme to assess theperformance of the trials has includedmeasurements of:-

i) Moisture contents of the road base,subgrade, shoulders and open ground ad-jacent to the trial sections. Themeasurements were made by test pitting orby profiling to a depth of 1.5 m belowthe surface.ii) Pavement strength in terms of CBR,using a dynamic cone penetrometer, to adepth of 0.8 m below the surface (Kleynand Savage 1982). Some insitu CBRmeasurements have been made which confirm-ed the correlation between DCP and CBR.iii) Transient deflection tests using theBenkelman Beam and measurements of radiusof curvature using the TRRIL 'line ofinfluence' apparatus (Smith and Jones1982) to give information on the struc-tural strength.iv) Surface condition in the form ofsurface deformation by cross-sectionaland longitudinal levelling surveys, rutdepths and cracking. Longitudinal levelsurveys have also been made with a mod-ified. Abay-beam (Abaynayaka 1984) whichalso provided information on road rough-ness.v) Traffic counts by 12 hour manualclassified counts supported by automaticcounter readings.vi) Rainfall and climatic data obtainedfrom the meteorological station at Lodwar.Average conditions for the past 30 yearswere also obtained from Lodwar.

B A

I A I I

6. RESULTS OF THE MONITORING PROGRAMME

6.1 Moisture contents

Moisture content measurements of the road-base which were generally taken from thecentre-line and edges of the pavementshowed that Sections A and B have dried tothe same value as Section C which had re-mained unchanged since construction. Thesubgrade moisture contents have remainedconstant at between 1 and 2 per cent to adepth of at least 1.3 m below the top ofthe subgrade. Moisture contents measuredin open ground have varied seasonally frommaximum values of 25 per cent at depth of0.8 m in the wet season to 2 per cent inthe dry season. The largest seasonalfluctuations occur at a depth of one metreas shown in Fig 14. They are howeverstrongly influenced by local variations intopography.

0

0.2 I-

0.4 f-

0.6 I-

0.8 J.0.a

0)1.0 ~-

Moisture content (per cent)10 20 30

1.2 [-

1.4 F

1.6

Fig. 14 Open ground moisture content profile

6.2 Pavement and subgrade strength

6.2.1 Subgrade

The median values of a series of DCP meas-urements showed that the CBR of Section Awas 59 per cent, the CBR of Sections B and

C were 29 per cent and 31 per centrespectively. In considering the sub-grade support strength using the modifiedstructural number approach to assesspavement strength (Hodges et al 1975) thedifferences for subgrades with CBR greaterthan 30 per cent are small. This can beillustrated by considering the designstructural number for the road which, in-cluding the contribution from the sub-grade, is 3.0. The proportion of thisvalue contributed by the subgrade is 63per cent and 64 per cent for Sections Band C respectively and 69 per cent forSection A. The modified structuralnumber values obtained from actual meas-urements are dealt with later in thediscussion of results.

6.2.2 Road base

Measurements in terms of CBR from the DCPtest are given in Table 4. They show thatSection A had the highest strengths andthat they had increased since constructionfrom a mean value of 61 per cent togreater than 100 per cent. This increasecoincided with the section drying fromthe optimum moisture content to 2 percent. Sections B and C, hav e remainedconstant since construction. The resultsof Section A correlate well with predic-tions from is0-CBR lines obtained fromthe laboratory tests (see Fig 11). Thefact that Section C did not correlate withpredicted values is discussed later inthe paper.

6.3 Pavement deflection and radius ofcurvature

Pavement deflections were measured in theverge-side and offside wheeltracks ofboth lanes at 10 chainages within eachtrial section. The results of two seriesof tests, one shortly after constructionand the other two years later, are givenin Table 5 and show that generally therewas no change between consecutive meas-urements apart from Section B in the lanetowards Lodwar which had increased.Section A had the lowest deflections andthe other two sections were approximatelythe same.

Whereas deflection measurements giveinformation related to the whole pavementstructure the radius of curvature, bymeasuring the shape of the deflectionbowl, is known to be more discriminatingin terms of road base behaviour. Oneseries of tests was carried out two yearsafter construction at the same locations

to

WetDryseason

1 1

Table 4. In-Situ CBR (DCP Method) for

roadbases

Section Year CBR per cent

No. of Men S RagValues Men S Rae

A 0 12 61 8 51-79

2. 3 79 1)14 1)63-892 14 166 114 70-450

B 0 12 40 10 21-551 3 45 4 41-472 14 41 9 29-60

c 0 9 37 6 27-441 3 37 3 34-392 14 47 8 32-61

(1) Inferredvalues of CBR greater than

100% exceed the calibrated range of DCPand are regarded as estimates.

Table 5. Mean Deflections

Section Year Transient Deflection(mm x 10-2)

Towards TowardsMarich Pass Lodwar

V/S 0/S 0/S V/S

A 0 34 31 37 452 33 32 34 40

B 0 42 48 49 512 45 48 58 63

c 0 44 48 50 552 45 48 49 53

as per deflection tests.given in Table 6 indicate

The resultsthat as for the

deflection tests Section A is the strongerwith the highest radius of curvaturevalues. Another reason for measuringradius of curvature and deflection of thepavement is to analyse the simple twolayer system of the pavement structureusing multi-layer elastic theory. Fig 15shows the relation between radius of curv-ature and deflection based on differentmoduli of elasticity for the two layers.E1 represents the road base and E2 the

subgrade. Plotted on the graph are the

mean values of each of the trial sectionsthat were measured two years after con-struction. The results show that themodulus of elasticity for the subgrade ishigh for all sections and that Section Ais different from Sections B and C. AlsoSection A originally compacted at optimummoisture content has the highest value of

E1. These differences between the layers

of each section and between the sectionsare consistent with those obtained fromother measurements such as the directstrengths of the pavement layers.

Table 6. Mean radii of curvature

Section Year Radius of Curvature (in)

Towards TowardsMarich Pass Lodwar

V/S 0/S 0/S V/S

A 2 80 83 80 66

B 2 40 36 44 42

c 2 50 46 52 43

.This method of structural analysis hasshown that it should be possible todistinguish between the behaviour of thetwo component layers in each of the threesections as historical data is built up.

6.4 Surface condition

Measurements of rut depth using a 2 mstraightedge, deformation from cross-section and longitudinal section profilesand inspections for cracking have shownno deterioration of the road surfacesince construction. The rut depths areshown in Table 7. The higher values re-corded in the offside wheel-track of theMarich Pass lane which persist throughall 3 sections were the result of a smallridge caused during construction by anoverlap of the surface dressing close tothe centre line. The use of the Abay beamfor measuring surface irregularity in thewheel-tracks has enabled road roughnessvalues to be obtained using a correlationbetween Abay beam measurements(Abaynayaka 1984) and the Towed FifthWheel Bump Integrator. The values whichwere between 2,000 and 2,500 mm/km areconsistent with normal results for a newroad with a granular base and sealed witha surface dressing.

t l

10 20 30 40 50 60 70 8090100Deflection x 10-2 (mm)

Fig. 15 Relationship between radius of curvature and deflection

6.5 Traffic volume and traffic loading

Manual classified traffic counts havebeen conducted annually since construc-tion. The results of these surveys havebeen supported by. automatic trafficcounter readings and have shown that theADT is 72 with 52 per cent in the mediumgoods and 3 per cent in the heavy goodscategory. Axle loads have not beenmeasured but using the Kenya Design Manual

guide of one equivalent standard axle pervehicl~e for medium goods type and foure.s.a for heavy, goods vehicles it is es-timated that a total of 20,000 e.s.a. havebeen carried in each direction in the twoyears since construction.

7. DISCUSSION

Quality control of compaction during roadconstruction requires that certainspecifications are met. In Kenya, like

17-

,0'

Table 7. Mean rut depths

Section Year .Mean Rut Depth (mm)

Towards Towards

Marich Pass Lodwar

v/S O/S 0/s V/S

A 0 3 6 4 21 2 5 4 22 3 6 2 2

B 0 2 4 0 11 2 5 2 32 3 6 1 1

c 0 2 8 3 21 2 6 2 22 4 5 1 3

many other countries, the requirement torroad bases is related to the laboratory4.5 kg rammer test. (Kenya specifies arelative density of 95 per cent of themaximum dry density at optimum moisturecontent). When compaction is carried outat low moisture contents it is doubtfulwhether the existing specifications areappropriate. This discussion thereforeconsiders the results of the study andhow they might be applied to enable drycompaction to be used more widely in aridareas.

7.1 Laboratory compaction at lowmoisture contents

Both dynamic rammer and vibrating hammermethods were used in the laboratory com-paction tests. At'low moisture contentsmore variable and lower densities wereobtained in the dynamic rammer testbecause the action of the falling rammeron the unconfined surface in the mouldwas to disturb the soil as well as com-pact it. In the vibrating hammer testthe soil is confined by the hammer's footwhich is just less than the diameter ofthe mould and more consistently highdensities were obtained. It is for thisreason that the vibrating hammer shouldbe the preferred method of laboratorycompaction at low moisture contents.

7.2 Comparison of laboratory and fieldcompaction

The overall comparison of laboratory and

field densities is shown in Figs 6 and 11.Fig 6 gives the results of the pilot-scaletrial and Fig 11 the full-scale trial.The difference between the laboratoryresults in the two figures is mainlybecause the soils used were from twodifferent sources although they were bothclassified as well-graded plastic gravels.The comparison of laboratory and fielddensities showed that at the lowestmoisture content of 2 per cent in thepilot-scale and full-scale trials relativedensities of 97 per cent and 100 per centrespectively were obtained. Thespecification density, therefore, wasreadily achieved and exceeded. In roadconstruction practice at low moisturecontents it is important to achieve thehighest density possible to ensure thatno further compaction subsequently occursunder the action of traffic. Specifica-tions related to the vibrating hammertest therefore, may be better for com-paction at low moisture contents or alter-natively field compaction trials to assessthe performance of the plant should becarried out to determine a 'method'specification.

At the intermediate moistdre contentthe relative densities achieved in thetwo trials were 89 per cent in the pilot-scale and 96 per cent in the full-scale.In the full-scale trial, therefore, thedensity passed the specification despitebeing in the moisture content zone wheredensities are normally expected to be attheir lowest. This was probably becausethe material used in the full-scale trialwas more rounded and would be expected tobe easier to compact. Although thelaboratory vibrating hammer test predictedthe trend of the changing densities be-tween low and intermediate moisturecontents it further confirmed the needfor compaction trials to ensure thatmaximum density is achieved in construc-tion practice.

At optimum moisture content the relativefield densities, as expected, exceededthe specified density, being 96 per centand 99 per cent for the full-scale andpilot scale trials respectively. Thedifference in these results was probablybecause of differences in the two mate-rials and in the two vibrating rollersthat were used. The densities achieved,therefore, would be dependent on thefield moisture contents and number ofpasses of the roller.

7.3 Laboratory and field CBR

The CBR results of Section A at construct-ion compared well with those predicted inthe laboratory tests. Because the roadbase has dried from 6.7 per cent to 2 percent since construction an increase instrength occurred which was also similarto that predicted (see Fig 11). SectionB has shown no change in strength mainlybecause the moisture content has changedmuch less, from 3.5 per cent to 2 percent. Section C has also not changed instrength since construction which is asexpected as the moisture content has notchanged. The field CBR of Section Chowever was significantly different fromthat predicted from the laboratory withvalues of the order of 40 per centmeasured in the field and greater than100 per cent predicted in the laboratory.The higher laboratory value is partlybecause of the lateral confinementprovided by the mould in the laboratorytest which is not present in the field.Studies elsewhere have shown that at lowmoisture contents the low strengthsmeasured at the completion of construc-tion of a layer have subsequently in-creased when confining pressures havebeen applied from the construction ofoverlying layers. This did not occur inthe full-scale trial with the road basecovered only by a thin surface dressing.

The higher strength of Section A bothat construction and during the monitoringperiod is considered to have been becauseof the type of material and the effectof moisture. The clayey gravel road basecompacted at the optimum moisture contentand then dried produced a very hard mate-rial because increasing suctional forcesbind the soil particles together. Thiscould only have occurred to a limitedextent in the 'intermediate' section(Section B) and not at all in the drysection (Section C) where the plasticfines in the dry condition only behavedas an inert filler. In the dry section,therefore., frictional forces aloneprovided the stability whereas at theother extreme, in~the wetter condition,strength was attributed to cohesive forcesand then increasing suction as the soildried as well as frictional forces. Itshould be noted that soil suction wouldnot normally be a contributory factor inroad bases because specifications prohibitor severely limit the amount of plasticfines that are allowed.

The results of the trials so far supportthe Kenya Design Manual which permitshigher limits of plastic fines for con-struction in arid areas. If normalspecifications were applied the additionof lime could be used to modify theplasticity of the soil. In such a casethe behaviour would be similar to thatfound in Section C where inter-particlefriction is primarily responsible for thesoil strength.

7.4 Field strength measurements and pave-ment design

Values of pavement deflection are consid-ered to be satisfactory for the type ofconstruction used although Section A wassignificantly different from Sections Band C. The difference between the sec-tions is also shown in the radius ofcurvature measurements where Section A hasthe highest and most satisfactory value.Because the difference is shown up in bothdeflection and radius of curvature resultsit can be deduced that this is because ofdifferences in the performance of the roadbase and not the subgrade.

Modified structural number can also beused to demonstrate that the pavementstrength is adequate. When the method ofanalysis is applied to the designrecommendations in Road Mote 31 (TRRL 1977)a structural number of 3.0 can be cal-culated and considered adequatefor a pavement designed to carry0.5 million equivalent standard axles.The calculated values for the trial sec-tions are 3.7 for Section A and 3.1 forSections B and C. This demonstratesfurther that all of the sections should besufficiently strong for the designedtraffic loading. However, a major featureof the monitoring programme is to examinethe performance of the road bases. Therelatively low strengths of the bases inSections B and C might indicate a risk ofshear failure under the action of trafficand if this does occur it would be expect-ed to be the result of a few excessivelyheavy axle loads rather than the cumula-.tive effect of traffic within the legallimits.

7.5 Construction methods

In the trials conventional methods ofusing construction plant were generallysuccessful when dealing with materials atlow moisture content. Segregation ofparticles is normally a problem whenhandling dry materials but in this case

14

spreading was easily carried out with amotor grader. The reason was mainlybecause the maximum size is 20 mm and not40 mm which is more common for crushedstone base materials.

Compaction of the dry material wascarried out with medium weight vibratingrollers. A method of rolling from theouter edges of the layer towards thecentre was adopted to give better confine-ment of the material. The pilot scaletrial indicated that other types of rollercould be used for compaction but it wasconcluded that vibrating rollers were moreefficient throughout the range of moisturecontents.

Considerable difficulties occur whenmixing very dry materials with waterespecially in hot dry climates when allow-ance has to be made for high losses caused..by evaporation. The quantities of waterrequired in conventional constructionpractice to raise the moisture content tothe optimum for compaction, when it maybe scarce and involve long haul distances,can seriously affect productivity.

7.6 Mechanism of compaction

The results of the study have shown clear-ly that for the well-graded clayey gravelhigh densities were obtained at lowmoisture contents. The mechanism toexplain this behaviour is still being in-vestigated but in normal compaction theaddition of water to lubricate soil part-icles to achieve higher density is wellunderstood; so is the effect of excesswater when saturation is reached when,jndividulal particles are ouished apart.With reference to Fig 2 this covers thepart of the graph in the dry density-moisture content relation represented byBC and CD. As moisture contents fallbelow the value of B in the figure, in-reases in density occur probably becausesurface tension between particles is re-duced as the soil approaches the drycondition. A further explanation of theincrease in density as soils become drierfrom B to A is that with compaction soilparticles will occupy the volume thatwould have been occupied by water in awetter condition and the high specificgravity of the soil particles comparedwith water will obviously result in ahigher overall density. It could also beexpected that at constant compactiveeffort closer packing of soil particlescan occur in the dry condition by the re-orientation encouraged by vibrating com-

paction.

8. CONCLUSIONS

The main conclusions arising from the studyof compaction at low moisture content were:

1. The vibrating hammer test is the pre-ferred method of carrying out laboratorycompaction at low moisture contents.2. High densitites were obtained at lowmoisture contents in both laboratory andfield compaction and could exceed currentspecifications.3. The vibrating hammer test predictedthe performance of the vibratory rollersover a range of low moisture contents.4. Compaction trials were important toestablish a method of compaction that willensure that maximum density is achievedin the field.5. For the dry compacted material in thefield trials the in-situ CBR resultsderived from DCP tests were much lowerthan those predicted by laboratory tests.6. The use of deflection and radius ofcurvature measurements and the determina-tion of the elastic moduli of the pavementlayers appears to provide a satisfactorymethod of monitoring trial sections andcomparing the performance of dry compac-ted road bases with those constructedconventionally.7. Construction plant and conventionaloperating techniques were successfullyapplied when using materials at lowmoisture contents.8. Road trials constructed with bases atdifferent initial moisture contents haveperformed equally well for two yearsalthough traffic levels have been low(20,000 e.s.a.). Further information onthe longer term performance is clearlyrequired before final recommendations canbe made.9. There is a need to extend studies ofdry compaction to a wider range of mate-rials used in different parts of the roadpavements and to determine the limitingclimatic conditions where dry compactioncan be used.

19... ACKNOWLEDGEMENTS

The work described in this paper formspart of the joint research programme ofthe Ministry of Transport and Communica-tions, Kenya and the Overseas Unit of theTransport and Road Research Laboratoryand is contributed by permission of theDirector TRRL and the Chief Engineer(Roads and Aerodromes) MOTC. The authorswish to thank members of the team who

Soil compaction at low moisture contents in dry areas in Kenya

participated in the work. Any viewsexpressed are not necessarily those ofthe British Government's Department ofTransport, nor the Overseas DevelopmentAdministration.

10. REFERENCES

Abaynayaka, S.W. 1984. Calibrating andstandardising road roughness measure-ments made with response type in-struments. In: Llecole Nationale desPonts et Chaussees. Roads and Develop-ment. Proceedings of the InternationalConference, Paris, 22-25 May 1984.Paris: Presses d'e L'ecole Nationale desPonts et Chaussees, 13-18.

BSI, 1981. Code of practice for siteinvestigation. 85 5930: 1981. London:British Standards Institution.

BSI, 1975. Methods of test for soils forcivil engineering purposes. 85 1377:April 1975. London: British StandardsInstitution.

Clegg, B. 1976. An impact testing devicefor insitu base course evaluation. In:ARRB. Proc. of the 8th Conference ofARRB, Perth 1976. Vermont South:Australian Road Research Board.

Department of Transport, 1986. Specifica-tion for highway works, Part 3. London:Her Majesty's Stationery Office.

Hodges, J.W._J.Rolt and T.E. Jones 1975.The Kenya road transport cost study:research on road deterioration.Laboratory Report 673. Crowthorne:Transport and Road Research Laboratory.

Kleyn, E.G. and P.F. Savage 1982. Theapplication of the pavement DCP todetermine the bearing properties andperformance of road pavements. FirstInternational Conference on the BearingCapacity of Roads and Airfields 1982.Trondheim, Norway: Norwegian Instituteof Technology.

MOTC Roads Department 1981. Road DesignManual, Part 3, Materials and pavementdesign for new roads. Republic ofKenya: Ministry of Transport.

Smith, H.R. and C.R. Jones 1982. Earlyperformance of some experimentalbituminous overlays in KenyaLaboratory Report 1043. Crowthorne:Transport and Road Research Laboratory.

Thurmann-Moe, T. and H. Ruistuen 1983.Graded gravel seal (Otta surfacing).Proc. 3rd Intl. Conf. on Low VolumeRoads. Tempe: Transportation ResearchBoard.

Transport and Road Research Laboratory1977. A guide to the structural designof bituminous - surfaced roads intropical and sub-tropical countries.Road Note 31. London: HMSO, ThirdEdition.

(C) CROWN COPYRIGHT. Published bypermission of the Controller of HerBritannic Majesty's Stationery Office.

Sn 1 rrimnir-i~nr) n 1 lxW' rnni f' fi r,r --nt-- in i --- ' -'-- i V -- 16