TRANSPORT RESEARCH LABORATORY TITLE by Experimental use of cinder gravels on roads in Ethiopia .D Newill, R Robinson and Kassaye Akfilu Overseas Centre Transport Research Laboratory Crowthomne Berkshire United K~ingdom IA 11 9th Regional Conference for Africa on Soil Mechanics and Foundation Engineering/Lagos/September 1987 9th Regional Conference forAfrica on Soil Mechanics and Foundation Engineering/Lagos/September 1987 Experimental use of cinder gravels on roads in Ethiopia D.Newill & R.Robinson Transport and Road Research Laboratory, Crowthorne, Berkshire, UK Kassaye Aklilu Ethiopian Transport Construction Authority, Addis Ababa SUMMARY: A full scale experiment has been carried out in Ethiopia to examine the performance of volcanic cinder gravels as the surfacing material for unpaved roads and as the road base under bituminous surfaced roads. Compaction trials were carried out to determine the type of plant to be used and an experimental road comprising 20 different sections was then constructed. Six sections were left unsurfaced and were monitored for 28 months during which they carried approximately 140,000 vehicles. A bitumen surface was provided for the remaining 14 sections .and these carried 150-200 vpd for 7Y2 years giving a total of 440,000 esa in one direction. Monitoring was carried out by taking quantitative measurements of. the performance of the road pavement throughout this period. As a result of the study, recommenda- .tions are made for the use of cinders in both paved and unpaved roads.' For unpaved roads, recommendations are made for a particle size distribution which provides a road surface that is resistant to corrugations. Improved performance can be obtained by mechanically stabilising cinders with plastic fines. *For paved roads, it is concluded that the types of materials used in this experiment are all capable of carrying in excess of 400,000 esa when sealed with a surface dressing and designed according to Road Note 31. Roadmixed asphalt is not a suitable surfacing for cinder gravels. In addition to the cinders, other materials also performed satisfactorily including a dry bound macadam, an agglomerate and a tuff. Cinders are easier to compact when they are mechanically stabilised with 10 per cent of volcanic ash soil. 1 INTRODUCTION The Joint Road Research Project of the Ethiopian Transport Construction Authority and the Transport and Road Research Laboratory (UK) has been studying various lowgrade materials to see if effective specifications can be devised for their use on lightly-trafficked roads. The work began with an investigation of volcanic cinder gravel which, although widespread in.Ethiopia, had only occasionally been used for road construction. The reason for its limited use was mainly because these gravels tend to be deficient in fine material and do not conform to the generally accepted grading specifications. In addition, they were reportedly difficult to compact. .';The study, aimed at a full examination of cinder gravels, comprised the following s~tages: C)A field survey to locate and identify cinder gravels and to obtain samples for laboratory testing; the survey also included the examination of aerial photographs and photo-mosaics and the preparation of a map showing the distribution of cinder deposits throughout Ethiopia. (ii) A laboratory investigation to determine their physical and engineering properties. (iii) An examination of existing cinder gravel roads to compare the performance of the materials under trafficking with the results of the laboratory invest~igations. (-iv.) -Compaction trials to determine the most suitable plant for practical use.: (v) A full-scale experiment to examine the behaviour of '.different cinder gravels under controlled conditions in relation to traffic and climate. The first three stages of the study were discussed in an earlier paper (Newill and Kassaye 1980) and the final two stages are described here. Reference was also made to other studies 467 of cinder gravels and principally those carried out in the USA (Hendrickson and Lund). In the compaction trials four cinders -from different sources were examined with and without plastic soil fines added and with and without water added. The full-scale experiment consisted of 20 sections of road built on a standard sub-base and subgrade. Each.section had a road base of different types of volcanic material with the exception of one crushed stone base used for control. Eleven sections were surfaced with a double surface dressing, three were surfaced with 50 mm of roadmixed asphalt and the remaining six sections were unsurfaced. The unsurfaced sections were monitored for two -years after which tine they were surface dressed. After one year, several of the unsurfaced sections were reconstructed. The surfaced sections were monitored for 7Y2 years. During this time, one of the~road mixed sections..was surfaced dressed, but no other pavement maintenance was carried out. 2 DEFINITION OF VOLCANIC CINDER Volcanic cinders are pyroclastic materials associated with recent volcanic activity. They occur in characteristically straightsided cone-shaped hills which frequently have large concave depressions in their tops or sides where mixtures of solids and gases were released during the formation of the cone. Cinders vary in colour, often within the same cone and may be red, brown, grey or black. The cinder particles also vary in size from large irregularly shaped lumps 0.5 m in diameter to sand and silt sizes. In some cones, however, particles may be more uniform with the largest size not exceeding 30 mm in diameter. Other characteristic features of cinders are their light weight, their rough vesicular surface and their high porosity. Usually they are weak enough to be crushed under the heel. An advantage as a road construction material ils the relative ease with which they can be dug from the quarry; a mechanical shovel or hand tools are usually adequate for their extraction although,occasionally, a bulldozer may be required to open up a working face. 3 PRELIMINARY INVESTIGATIONS The main conclusions from the preliminary investigation of cinder gravels which covered a field survey, a laboratory study and an examination of a cinder gravel road, are given below (Newill and Kassaye 1980):- (i) Cinder gravels are more widespread in Ethiopia than was originally believed; this showed the value of using aerial photographs in survey work and enabled a preliminary nap to be prepared giving the distribution of cinder cones. (ii) In order to obtain representative material from a cinder cone, it is important that samples are taken from below the weathered zone .which can extend to a depth of two metres. .(iii) Although 'as dug' cinder gravels do not meet the recommended grading requirements for road base materials, the laboratory investigation revealed that, because of the weak nature of the aggregate particles, breakdown under compaction occurred with an improvement in both grading and strength properties. (iv) In the laboratory investigation, the cinder gravels were not affected by changes in moisture and even complete immersion in water only reduced their strength slightly. (v) The addition of locally available plastic volcanic ash soil, to make up for the deficiency of fine material in the grading, improved the mechanical stability of cinder gravels and indicated that this could be a valuable construction practice. However, unlike the natural cinders the mixed materials lost some of their strength when they were saturated with water. (vi) The gravel road study confirmed that an improvement in the grading and the strength of cinder gravels occurred under normal road conditions even when trafficking was used as the means of compaction. The results from the preliminary investigations indicated that cinders cotuld provide useful road construction materials especially for gravel roads. However, it was necessary to carry out further work under known conditions of traffic and climate in bituminous surfaced roads, as well as in gravel roads, before limits could be recommended for their various uses. It was therefore decided to construct pilot scale compaction trials and then a full-scale road experiment to examine these aspects further. 4 COMPACTIO0W, TRIALS 4.1 Organisation of the trials Because doubts had been expressed about the ability to compact cinder gravels in normal road construction practice, a series of pilot-scale compaction trials were under468 taken before carrying out full-scale road experiments. Earlier laboratory tests had shown that, for typical cinder gravels, repeated compaction caused the breakdown of cinder particles to give an improved grading and an increase in density. An improvement in grading and better compaction could also be obtained by mechanical stabilisation with the volcanic ash soils which are normally found close to cinder deposits. It was also shown that, unless subjected to repeated compaction, the densities obtained were not affected by changes in moisture content. For these reasons, the compaction trials were designed to examine the following conditions: (i) crushed and uncrushed materials. (ii) with add without fines added. (iii) with and without water added. Four different cinder gravels were used in the trials. of these, two, one from Mojo and one from Bekojo, were classed as typical cinders, but the strengths of their cinder particles were different. Another, from Nazret, had relatively uniform sized cinder particles, whereas the fourth, from km 130 on the Awash Malkassa to Assela construction project, was unusually a well-graded material with a higher proportion of fine-sized particles. Description and properties of the materials are shown in Table 1 and Fig 1. The volcanic ash soil used for mechanical stabilisation was a clayey silt with a plasticity index between 8 and 12. Table 1. Cinders used in compaction trials Cinder ~'ael Moiidar- - soregte` Jrt Descripticn value - per cent IMbjo 1.40 Typical black ci~~e Bekojo 93 Typical red cinder Nazret 90Single-sized cinder mn 130 59 Well-~de Awash-Assela. Road cinder Different types of compaction plant were used in the trials. They comprised a Galeon Roll-o-Static 10 tonne smoothwheeled roller, a Bomag BW 200 7 tonne vibrating roller and a Hyster C530A pneumatic-tyred 10 tonne roller. They were used singly and in different combinations BRITISH STANDARD SIEVE SIZES I Mn .Im nmm o " 0 a) 0 ) -. OM N 0 = 10 ID In I * 0 flW RO - N, Ni W.11 - N lo0 L. . ...J 4. . .. I III~~J 1V 80 INAZRE 70 i-- 1 6(1 ....L - - I I _ 0 _ _ - 4(1 ___ a~~~~~J.4o Sand Gae F i.e Medinum Coa,. FiM. uMdi. C~ Fig.1 Particle size distributions of cinders used in comnpaction trials to determine the most effective way of compacting the cinders. The site for the trials was at the Mojo cinder quarry on a cleared area of the quarry floor. The other materials were imported and stock-piled at the site. For the comparison between crushed and uncrushed cinders sufficient quantities of crushed material were prepared by spreadi~ng thin layers of 100 mm of cinder on a specially constructed concrete slab and compacting with the vibrating roller. The individual pilot-scale compaction trials were carried out on the materials laid in loose layers 200 mm thick in sections 15 m long and 3 m wide. After compaction, the densities were measured using the sand replacement density test and the means of three tests for each trial are those given in Table 2 where they can be compared with the maximum densities obtained in the standard laboratory compaction tests (BSI 1975). 4.2 Results of the compaction trials The compaction trials showed that:- (i) Of the four cinders used, the Mojo and Bekojo cinders behaved similarly and 469 1 n 5 1 a1 h1 :2 1 Table 2. Compaction trials on cinder gravels with different treatments Cinder - as dug Cinder + water Cinder + fines + water Type of Type of ca~pticr Density r'g/n? cinder ULkicnhe Crushed thcruhed Crused Ukrcrx~ Chushed Vibratory roller, 12 passes 0.99 1.22 - - - - ?vflTO Pneuzratic tyred, 2 passes + 1.09 1.12 1.09 1.17 1.07 1.26 smooth-~~eed, 8 passes Pneumatic tyred, 2 passes + 1.12 1.27 1.12 1.30 1.11 1.26 smooth ~,ieled, 16 passes labratory 2.5 kg ra~~ 1.03 1.25 1.11 ca ticn 4. 5 kg ram~ 1.30 1.63 1.30 Vibratory roller, 12 passes 1.09 1.15 - 1.23 - BEK(J0 PMR, 2 passes + 1.12 1.31 1.26 1.36 1.12 1.30 smooth~-~ieed, 8 passes PMR, 2 passes + 1.17 1.35 1.35 1.39 1.17 1.33 smooth~~e ed, 16 passes Labnratory 2.5 kg rair- - 1.22 1.38 - - carpactio 4.5 kg ranner 1.44 1.73 Vibratory roller, 12 passes -- 1.25r 1.38B Vinl30 PMR, 2passes + -- 1.28 1.35-- smnooth-~keed, 8 passes PTR, 2 passes + 1.35 1.36 smooth-~eled, 16 passes Laboratory 2.5 kg rame -- 1.43 --- caiation 4.5 kg rarre 1.49 the other two behaved very differently. (ii) The relatively uniform-sized Nazret cinder could not be compacted satisfactorily with any of the rollers in the crushed or as-dug condition, with or without water added. Although compaction could be achieved with the mechanically stabilised mixture the Nazret cinder was ,regarded as unsuitable for consideration as a road base material and therefore no test results are reported. C iii) The well-graded km 130 cinder was little changed by the preliminary crushing and had sufficient fines to give a tight well-knit finish in the compacted layer. It was not necessary to add further fines for mechanical stabilisation and only the compaction results with water added are reported. Civ) For the Mojo and Bekojo cinders in their natural condition, it was necessary to start compaction with the pneumatic tyred roller working from the outside of the section towards the inside in order to prevent the materials being displaced sideways. With the vibrating roller, which was lighter than the smooth-wheeled roller, the first two passes were made without vibration. (v) For the Mojo and Bekojo cinders, higher densities were obtained with the crushed material than with the uncrushed. Densities of the as-dug dry materials were not noticeably changed either by the addition of water or the addition of fines. However, the addition of fines made compaction easier and produced a tightly-bound surface which could not easily be scuffed or damaged by brushing with a broom. The addition of water also gave a better surface finish especially as the material dried. Cvi) In the compaction trials, the densities achieved with the combination of the pneumatic-tyred and smooth-wheeled roller were generally higher than with the 470 vibrating roller. This was probably because of the greater mass and higher energy input of the roller combination. It is worth noting that the compaction plant-used in the trials was that being used in the construction of the Awash Mvelkassato Assela road. It was lighter than that often used in road works and it is likely that better results could be obtained with heavier plant. Cvii) The highest densities achieved in the trials were generally close to those obtained in the standard 2.5 kg rammer laboratory test. It should be noted, however,,that for typical cinders like Mojo and Bekojo there was a large difference in the maximum densities between the 2.5 kg and the 4.5 kg rammer tests. This means that normal specifications for compaction of road base materials, which are usually related to the heavier laboratory compaction test, are unlikely to be achieved. Clearly, this is because greater breakdown of particles occurs in the laboratory test which is conducted in a confined mould. An alternative specification would therefore be required for cinder gravels which would have to consider the relation between density and breakdown of particles. It is recommended that, for typical cinder gravels used in road construction, compaction trials should first be carried out to establish the method of compaction and the target density to be achieved. If a 'method' type specification is adopted, this would also help to reduce the number of difficult in-situ density tests that have to be made. It was concluded, therefore, that typical naturally-occurring cinder gravels can be compacted in practice, but care is needed in selecting the most suitable plant. For the full-scale experiments described in this paper, it was recommended that 2 passes of a pneumatic-tyred roller followed by 12 passes of a smooth-wheeled roller should be used. 5 FULL-SCALE EXPERIMENT 5.1 Objectives of the experiment A site for the full-scale experiment was offered by the Ethiopian Transport Construction Authority on the Awash MelkassaAssela road. This 80 km long gravel road, consisting of an old three metre wide telford base which had subsequently been widened to six metres, was being upgraded to bitumen surfaced standard by complete reconstruction on the old alignment. The whole length of road was traversed and subgrade investigations were carried out at two possible exper imental sites. The one eventually chosen was six kilometres south of the town of Dhera where the main camp of the construction project was situated. The objective of the experiment was to compare the performance of a variety of cinder gravels under the same conditions of subgrade, sub-base, climate and traffic to assess the most effective way that they could be used asroad building materials. To achieve this, the cinders were laid as a gravel wearing course in addition to being used as base material under a surface dressing and under a 50 mm roadmix surfacing. The cinders were laid as 'pit-run' material and were also mechanically stabilised with a volcanic ash soil of a clayey silt texture which was the natural subgrade material in the area of the experiment. On the gravel sections, both compacted and uncompacted cinder gravels were laid. In two of the sections constructed with a surface dressing, the 'as dug' cinders were initially passed through a crushing plant to produce an imfproved grading before compaction was carried out. The roadmix surfacing was used because it was thought that based constructed with 'as dug' cinder gravels might not have sufficient cohesion to withstand deformation by trafficking if only a relatively thin surface dressing was used. Duplicate sections were constructed, therefore, to. compare the performance of cinder gravel bases with two types of surfacing. Subsidiary experiments examined the performance of other volcanic materials. 5.2 Pavement design The soil at the experimental site was a uniform fine volcanic ash with a plasticity index between 8 and 15. This gave a subgrade with a soaked CBR of 7 per cent. To obtain uniform subgrade conditions over the whole of the experiment, it was decided to remove the existing telford base, and this was done with a bulldozer and a grader. A preliminary traffic count estimated that, of a total ADTof abouft200, 80 to 100 commercial vehicles per day were using the road in both directions, and an axle-load survey already carried out by the Joint Research Project~on other roads indicated that the number of equivalent standard axles per commercial vehicle for Ethiopia was 2.33. Because of the economic situation in the country, a traffic growth rate of zero was assumed. Thus, for a 15 year design life, it was predicted that the road would carry: 471 10x 365 x 15 x 2.33 = 0.64 mi 2 equivalent standard axles. The basic pavement design of th ment was calculated using Road No (Transport and Road Research Labo 1977). The cumulative axle loadi with a 'soaked' subgrade CBR of 7 gave a design thickness of 150 mm with 150 mm of sub-base. The thi the experimental construction was uniform throughout, irrespective different kinds of material being each section. The design thickness of the exp sections contrasted with the main which was 200 mm of sub-base laid layers with 150 mm of crushed roc 5.3 Layout of experimental sectio The cross-sectional profiles were same as those on the main contrac sub-base width of 10 metres cover S metre base and 2 metre wide sho (Fig 2). The base width was incr 8 metres on the unsurfaced sectio the shoulders reduced to one metr ~~ Orn- Formation-. -.---- 7m- Primed- -.-- 6m-----Surfaced-. Sections 1 -14 Double surface dress .150mm base -150mm sub-base Sections 15-20 2.5 Slope 2:1 150mm ~gravel surfacing 1mmsub-base Fig.2 Cross-section profiles of the experimenta llion representing a wide range of particle llion strengths: Mojo, Bekojo and km 130 (see Table 1). The cinder from km 130 was also, e exeri- used as the common sub-base for the experite expri ment as it differed from any other cinder rteo31 sampled in being better graded and having ratory a much higher percentage of fines. The. ng coupled layout of the experimental sections is per cent shown in Table 3. of base The compaction trials indicated that it ckness of might prove difficult to produce a tightlykept knit surface on some of the sections that of the were subsequentl-y to be primed and surfused for aced. Fines were therefore added to most of these cinder sections to help provide erimental the necessary cohesion. Both clay fines, contract mixed during laying, and quarry dust, in two vibrated, into the surface after laying, k base. were used. One section was laid without the addition of fine material. Two other materials which did not meet ns the ETCA specifications for road bases were also included among the sections to kept the be surface dressed. These were a volcanic t with a tuff and a volcanic agglomerate. The ed by a former had been used as sub-base on the ulders Addis Ababa-Nazret road and the latter was eased to being used as~the sub-base material on the ns and main contract. To demonstrate a technique e. of construction for use in arid areas, a section of dry-bound macadam was included and laid in two 75 mm layers. This used a single-sized (50-60, mm) aggregate with quarry fines vibrated into the voids. A control section of crushed rock, as used Varabl slpe for project base, was also included. Varabl slpe Three of the cinder sections were also re peated under the roadmix surfacing. ing ~~Of the six sections constructed with gravel surfacings, four were laid in the same way as the base in the bitumensurfaced sections. The other two sections comprised, material laid 'as dug' with ~~> ~compaction being left to subsequent trafficking which was the form of construction used on the low-cost road schemes 1 secions inspected during the first stage of the Isecions project (Newill and Kassaye 1980). The experiment was divided into three parts: eleven 60 metre sections were surfaced with a double surface dressing using single sized chippings (BSI 1971); three sections, one 60 metres and two 120 metres in length, were surfaced with ETCA standard roadmix asphalt; six 100 metre sections were left unsurfaced as a gravel road. With traffic levels of the order of 200 vehicles per day, the experiment provided the opportunity for accelerated testing of unsurfaced materials. The major part of the experiment was to compare three different types of cinder 5.4 Materials properties Particle size distribution charts for the sub-base used in the experiment are shown in Fig 3 and charts for the materials used for road base under paved surfacings are shown in Figs 4-6. Gradings of the material used for gravelling the unsurfaced test sections are shown in Fig 7. Measurements of in-situ density and moisture content of the road base materials were recorded at the time of construction, as were in-situ subgrade CBR values. 472 Table 3. Experimental sections Length Section (metres) Surfacing Material Mean dry Moisture density content (mg/n3 ) Subgrade -- Volcanic ash 1.49 15.2 103 Sub-base - - Km 130 cinder 1.49 13.5 94 1 60 Km 130 cinder 1.53 14.4 97 2 60 Km 137 crushed stone 19 7.2 97 -' (Rhyolite used as base U) material on main contract) 3 60 . Dry-bound macadam - - 4 60 Sodere agglomerate 1.90 8.9 102 5 60 Mdojo cinder *i fines 1.49 9.1 103 S so Bekojo cinder + fines 1.43 10.4 101 7 60 Bekojo + quarry fines 1.41 10.2 100 8 60 Crushed Bekojo + fines 1.59 14.7 113 9 60 - Crushed it1ojo + fines 1,50 16.4 10.4 10 60 Mojo cinder 1.34 14.6 103 0 11 50 Nazret tuff 1.39 20.8 93 12 60 . Beko~jo cinder + fines 1.45 11.5 103 13 120X Modjo cinder + fines 1.45 13.8' 101 14 120 X E Km 130Ocinder 1.48 15.5 94 15 100 Km 130 cinder 1.51 14.8 96 16 100 Bekojo cinder. + fines 1.49 11.5 106 17 100 Bekojo cinder (not - - - compacted) 18 100 Mojo cinder--(not-... In compacted) 19 100 Mojo cinder +~ fines 1.45 12.4 101 20 100 Sodere agglomerate 1.91 10.6 J37 traffic counts Date of count Etm ed Percentage .(Time since Etm ed of buses construction) D and trucks 6 months 210 53 28 months 156 50 53/2 years 173 .54 6Y2 years 230 51 Table 5. Results of axle load surveys Mean equivalence, Date of survey factor per vehicle (Time since Towards Towards construction) Assela Nazret 6 months 1.3 2.5 28 months 1.2 2.9 6Y/2 years 2.2 3.6 These are recorded ments were made on in Table 3. No measureSection 3 which was of dry-bound macadam construction.' Sections 17 and 19 were originally constructed without compaction so the initial density measurements were deferred for three months until compaction by traffic had taken place. 5.5 Maintenance and reconstruction of sections One year after construction, three of the six gravel sections Nos.17, 18 and 19 were reconstructed. On two other sections, Nos 15,. and 16, the upper 50 to 100 mm of the material was loosened with a grader and respread, then watered, shaped and compacted. New material was provided for Sections.17-19 from the same quarries (Bekojo and'Mojo) that were used previously. Again, 10 per c ent of fines from adjacent to the experimental 473 Per cent compaction (BS heavy) Table 4. Results of BRITISH STANDARD SIEVE SIZES aD 0 N 0 BRITISH STANDARD SIEVE SIZES Lana, ----------------- mID .- I-: - .M No N W 0 aDQ - t- N ' . 2 ~ - loo 70 o~~~~~~~~~~~~~~Gae so Si3Gandigevopofsbae materil BRITISH STANDARD SIEVE SIZES at n-ON Q ~ N Cla m o N 90 a0 O Normal base grading envelopea 70a I0 aon 70 a I~~ 40 aa 30~~~~~~~~~~~~~~Gae 20 i m oas F9.4 Gradings of cinder gravels in surface dressed sections DO 1 1 1 1 ri 1 :A .F Ill action 1 80 : 3 70 4 Base J 60 Normal gradin enve lop It so 4 1 e X 30 A; 2 0 J11 iP 10 0 di.. Coarsa/ Fine Medi.,n tease Fig.5 Gradings of other base materials in surface-dressed sections BRITISH STANDARD SIEVE SIZES 90~~~~~~~~~~0. grading envelope~ ~~~~Daaee Fig.6 Gradings of bases under road-mix sections 474 0 N.~o
0 an R '
. 0
a !2
n 0 .- 2
0 1 1 1.1 a 1 ,a 1 ot c1 t 1 R z 91
R 1 2 z 1 U 1 It BRITISH STANDYARD SIEVE SIZES [.* - ,um 11. ~igm 100go' 10 0~~~~~~~ lr Secind 15an 1 16 >U ar1e IJIaL_ Fig.7 Gradings of gravel test sections construction area were added to the cinder of Secti on 19. These materials were watered, graded and rolled, thus providing a comparison with the previous construction where the compaction was carried out by trafficking on Sections 17 and 18. Section 20 was not reconstructed or maintained at this stage. During the first year of trafficking, maintenance grading was carried out twice: once on Sections 15-19 and once on Sections 17-19 only. No maintenance at all was carried out during the second phase of the experiment. After 28 months, monitoring of the gravel sections ceased. At this time, residual gravel was bladed off the sub-base which was then primed and sealed with a single surface dressing. At the time of the monitoring 28 months after construction, the surfacing of Section 13 was. badly cracked and it was necessary to take some action to prevent failure occurring during the next rainy season. As plant was on site to surface dress the gravel sections, the opportunity was taken to apply a single surface dressing to one half section to seal the cracks. This enabled future comparison to be made with the remaining half of the section which had not been resurfaced. Other than this, no pavement maintenance was carried out on any of the surfaced sections during the 7Y% years in which they were monitored. E~ 140- =120-U 100- E80 o 60- 40 20- 0 JJASO NDJFMAMJJASON DJ1FM AMJJA SO Time Fig.8 Monthly rainfall during first two years of experiment 6 CLIMATE Rainfall records were obtained from an automatic rainfall intensity gauge installed in Dhera village approximately 6 km from the experimental site. Back-up readings were obtained from a daily rainfall gauge operated by the Central Arussi Development Unit (CADU) at the same site. A histogram showing the monthly rainfall pattern during the first two years of the experiment is shown in Fig 8. It is known from national climatological data (Mesfin 1967) that the mean temperature at the site is approximately 190C and that this varies little throughout the year. The annual mean rainfall for the area is approximately 700 mm with the wettest months from May to September. 7 TRAFFIC SURVEYS AND AXLE LOADING During the course of the experiment, four traffic surveys were carried out. These comprised manual classified counts of vehicles for seven consecutive days. On three of these surveys, axle loads were also measured using a portable weighbridge located close to the site of the experiment. A summary of the surveys is given in Tables 4 and 5. ..Th~e, reducti~on in-flow between the 6 and 28 month surveys reflects the removal of a large amount of construction traffic from the road following the completion of the Awash-Assela project 11 months after construction of the experimental sections. It is clear from Table 5, that the Nazret traffic lane was the most heavily laden, and the detailed results from the survey show that many trucks travelling in this direction 475 W W1
,n
In ' ' -
2 5P cz !R 1 ,I; a 1 1 z 9
. 0 0 0 . 0
0 were loaded with agricultural produce, whereas there was a higher percentage of empty vehicles travelling in the other direction. In the absence of axle load data for the period between 28 months and 6Y2 years, the mean equivalence factor for intervening years has been taken as the mean of these two values.. Thus, values of 1.7 in the Assela direction and 3.3 in the Nazret direction have been assumed. On this basis, the cumulative axle loading in the heavier laden direction was determined to be 440,000 esa over the 7/2 year life of the project. per section was made and those were taken on the centre line and left and right hand lanes varying along the road. Readings were taken one year after construction and again a year and a half later and after 6 years. The moisture content of the subgrade at the points where CBR tests were carried out were also measured. In addition to the CBR measurements, a Clegg Impact Hammer (Clegg 1976) was used to determine subgrade strength during the second round of measurements. Pavement strength-was also assessed using a dynamic cone penetrometer (Kleyn and Savage 1982) during the final round of monitoring. 8 MONITORING OF PERFORMANCE 8.1 Density and moisture content In situ densities of the base were measured using the sand-replacement test with a 4 inch cone. Samples were weighed in the field then returned to the laboratory where moisture contents were determined. On the paved sections, three measurements were taken spread equally along the length of the road and arranged to cover the centre line and two metres on either side. On these sections, it was necessary to remove the surfacing in order to take the measurements. On the gravel sections, five equally spaced measurements were made in a similar way. Here, results were corrected for the irregularity of the surface. 8.2 Particle size distribution Grading samples were collected adjacent to each density hole. The samples for each section were combined and then analysed by wet sieve analysis. No gradings were carried out on the dry-bound macadam base of Section 3. With this exception, the paved sections were sampled on four occasions: immediately after construction and after 1 year, 2 years and 6 years. The gravel sections (15-20) were sampled at more frequent intervals during their life. 8.3 Subgrade strength In situ California Bearing Ratios of the subgrade were made in holes dug through the pavement layers using a rig mounted on the back of a truck. One measurement 8.4 Cross-section levelling and rutting The surface of all the sections were levelled at regular intervals. Readings were taken on a grid of 25 centimetres across the road and 10 metres along the road. Paved sections were marked with paint at 10 metre intervals along the centre line to act as a datum for the measurements and 15 centimetre nails were knocked into the gravel sections at 30 metre intervals along the centre line between which a measuring tape could be stretched. Elevations were tied to bench marks established approximately 30 metres either side of the road at the start and finish of each test section. Throughout the period of the experiment, elevations obtained were used to plot profiles so that the deformation of the surface could be monitored. Using these profiles, the formation of ruts was measured under a simulated two metre straightedge. During the later stages of the experiment on the paved sections, any rutting in the wheelpaths was measured directly under a 2 m straightedge. The elevations obtained at each survey from the gravel sections were read directly from field sheets into a computer program which-.determnined the loss of surfacing material between each survey. 8.5 Corrugations The formation of corrugations on the gravel sections was monitored both before and after reconstruction. Initially, corrugations were measured by spanning adjacent crests with a straightedge and 476 measuring the depth of the trough with a steel tape but, after six months, a new method was introduced and elevations of 15 crests and troughs were found using an engineer's level and staff. In both cases, measurements were made on the middle 30 metres of the section, traverses being carried out on the centre line and at one metre on either side. 8.6 Roughness An assessment was made of the roughness (riding quality) of the gravel sections using a vehicle-mounted bump integrator, measurements were corrected for variation of the vehicle characteristics by checking readingson a two kilometre long calibration strip. However, it was not possible to compare readings with those of a standard instrument, hence the values obtained are not absolute. Measurements were made during the period following reconstruction to the time the gravel sections were paved. 8.7 Deflection Deflection measurements were made on the paved sections of the experiment using deflection beans and the standard TRRL method (Smith and Jones 1980). Points where deflections were to be measured were painted on the road and all readings were taken within 150 mm of these. Measurements were made in both wheel tracks and in both traffic lanes at 5 metre intervals along the road in the centre 40 metres of each test section. Temperature calibrations were carried out and readings were corrected as necessary for road temperature. 8.8 Cracking Measurements of cracking were made by placing a one metre square on the road surface on the centre line and at two metres on either side at a point 27.5 metres from the start of each...section. Any cracks were marked with chalk and the area was photographed from vertically above. The resulting photographs were then projected on a screen at full size enabling the total length of cracking to be measured directly. 9 PERFORMANCE OF THE GRAVEL TEST SECTIONS 9.1 Monitoring The gravel test sections were monitored for a period of 28 months. Total traffic in this period was approximately 140,000 vehicles (70,000 in each lane) which was divided into two phases. The first phase lasted for 11 months following initial construction, after which Sections 15 and 16 were reshaped and recompacted, and Sections 17 and 19 were regravelled and compacted. During the-second phase, lasting for 17 months, no maintenance was carried out. Section 20 received no maintenance throughout the whole period of the experiment. The performance of the gravel test sections was assessed by measurements of deformation, by loss of material and by changes in materials properties. Deformation is indicated principally by the formation of ruts and corrugations, as well as general loss of shape and an increase in roughness. 9.2 Deformation 9.2.1 Rutting The relationship between rut depth and traffic is shown in Figs 9-13. During the first phase, the range of mean rut depths for the different sections was from 22 mm for Section 20 to 44 mm for Section 19, and this agreed fairly closely with measurements made in the second phase. It is important to note, however, that for all of the sections there was considerable variation in the rut depths of individual wheel-tracks. In general, the greatest rut depths occurred in the Assela-bound lane, which surprisingly was not the lane carrying the heavier traffic, although vehicles were climbing a slight gradient in this direction. The maximum mean rut depth measured in any wheel-track was 114 mm (Section 19, Assela lane, outside wheel-track) and it is clear that all of the surfacing mater- .ial was lost .from this wheel-track. On site, this was observed by the change of colour of the gravel as the sub-base showed through. Under non-experimental conditions it i!5 clear that maintenance by motor-grader would be required before this condition was reached. lack of compaction during construction was apparently not detrimental to the materials' ability to withstand the formation of ruts. 477 55 E .5 c0 C 0 10 20 30 40 50 60 70 Cumulative traffic volume (Thousand vehicles) Fig.9 Rutting of Section 15 50 451 401 .E 0. 351 301 251 201 151 101 5 80 0 10 20 30 40 50 60 70 80 Cumulative traffic volume (Thousand vehicles) Fig.10 Rutting of Section 16 55 9.2.2 Corrugations These formed on all sections except Section 20, although the degree of corrugation varied considerably. In all cases, corrugations were more severe on the Nazret-bound lane. Thus, the road tended to rut in the Asse la-bound lane where there was a slight gradient, but where axle loads are small, and the road tended to corrugate in the other lane ..hich has a slight downhill gradient and a higher axle loading. The reasons for this are likely to be that corrugations are dependent on vehicle speed (Heath and Robinson 1980) which would probably be higher on the downhill gradient, and that the abrasive action of' vehicles climbing the gradient could wear---ruts in the road. Because of the short length of the test sections and the relatively low level of traffic flow, it was not practical to measure vehicle speeds over the experimental sections in order to confirm this. It is thought that axle loading contributes to both rutting and the formation of corrugations and it is difficult to relate the differences in the performance in the two directions to the different E 0.a *0 'E 0 10 20 30 40 50 60 70 80 Cumulative traffic volume (Thousand vehicles) Fig. 1 1 Rutting of Section 17 478 - After grading - /~~~~ / / Original construction / - / / - / / E35 -- E 30 ~25 20 I O~~~~~~~~~riginal construction 15 (without compaction) 10 / 5, 0 10 20 30 40 50 60 70 81 Cumulative traffic volume (Thousand vehicles) Fig. 12 Rutting of Section 18 50 -After SECTION 19 reconstructio 240 E 30 30 ~~~~Original construction 10. 0 20 40 60 80 100 120 140 Cumulative traffic volume (T'housand vehicles) Fig.13 Rutting of Sections 19 and 20 mean equivalence factors. Sections 17 and 18.both formed severe corrugations within a few weeks of the initial construction. This severity increased through the dry season but then reduced with the onset of the rains. During this period, minor corrugations formed on Sections 15 and 16, but no -E 0 0 Dry Wet Dry Wet eaton season season season Section 18 Mojo cinder 'as dug' Section 16 ~ ' Bekojo cinder/ + 10% fines/ /' Section 1 7 Bekojo cinder 'as dug' -- - - Section 19 -* Mojo cinder / + 10% fines / 0 10 20 30 40 50 Traffic - Vehicles X 1 000 60 70 80 Fig.14 Depth of corrugations of gravel surfaced sections (Nazret Lane) corrugations formed on Section 20. Although corrugations did form on Section 19, these were much less severe than on Sections 17 and 18. The resistance to corrugation of the various sections was a major factor when designing the reconstruction after 11 months. Sections 15 and 16 which had performed relatively well were then reshaped, watered and rolled without the addition of any new material, whereas Sections 17 and 18 were completely reconstructed using new material from the same quarries as the original, but this time with compaction at optimum moisture content to try and improve the subsequent performance. Section 19 was completely reconstructed to the same specification as the original to provide a control section. Section 20 which had performed very well was left untouched. After reconstruction, Sections 17 and 18 again formed corrugations very quickly, but these never reached the severity which the uncompacted material had achieved. Again there was a tendency for the severity of the corrugations to cycle with the wet and dry seasons and an overall tendency for the severity of the corrugations to reduce with time as the surfacing material 60 was gradually worn away. Sections 16 and 19 did not exhibit corrugations until six months after reshaping and reconstruction respectively and these were of a very minor nature. Section 20 continued not to corrugate and, following reshaping, no corrugations appeared on Section 15 either. The corrugation measurements made during this second phase are shown in Fig 14. From these results, it is clear that the untreated cinders in Sections 17 and 18 are prone to the formation of corrugations whereas the materials in the other sections are relatively resistant to their formation. The grading envelopes of these two groups of materials at the time of construction are plotted in Fig 15. It is clear 479 100 CL 60 .c 40 20 .075 .150 .425 1.18 2.36 5.0 10.0 20.0 50.0 Particle size (mmn) Fig. 15 Grading envelopes of corrugation resistant and prone materials from this that the corrugation resistance of the~se cinder gravels is related to their fines content. Materials with a fines content of 10 per cent or less are prone to corrugations, and corrugation resistant materials have a fines content of more than 10 per cent. The volcanic ash soil added to the cinder had a plasticity index between 8 and 15 and the cohesion this provided contributed to the improved performance as a gravel wearing course. It is interesting to note that the grading envelope for corrugation resistant material was wider than that norsally recommended (TRRL ORN 2) but the plasticity index was similar to the preferred limits. 9.2.3 Roughness The progression of roughness is shown in Figs 16-18 and it can be seen that these readings correlate well with the cor-rugation observations. The rapid formation of corrugations in the Nazret-bound lane of Sections 17 and 18 after the reconstruction can be clearly seen. Sections 18 and 19 start off relatively smooth, and Section 16 exhibits characteristics midway between the two. As observed on site, the general level of roughness on Section 20 was considerably higher than the other sections. This is because of the amount of oversize material in this gravel. The riding quality of all sections deteriorated with time, but again the seasonal trend is observed with conditions improving during the wet seasons. The reason that all of the readings fell at the final measurement is not understood, but may be connected with a calibration problem experienced at t-hat time. Roughness values in excess of about E10 1 7 14 0 ~~~~~~Thousands of vehicles 20 40 60 80 Fig.16 Roughness progression for NAsselaBound Lane 16 -14 -sel-bun ln 13 1 12- ~~~~~~ ~ 11~ ~ ~~~~~~~~~1 10 0 19 8 60 8-10 12 Fig.17 Roughness progression for sazectioun 20n exeee orcm ls oti au evetualy 480 70 Em 02 _j 8g 8 0L E iJ 'U 0 0 _j 60 p. 50O* 30 20 1 0 0 80 70 60 s0 40 30 20 1 0 7 1 60 50 40 30 20 1 0 0 1.r 20 40 60 80 100 120 140 Thousands of vehicles Fig.19 Gravel loss on sections 15 and 20 20 40 60 80 100 120 140 Thousands of vehicles Fig. 20 Gravel loss on sections 16 and 17 20 40 60 80 100 120 140 Thousands of vehicles Fig. 21 Gravel loss on sections 18 and 19 9.3 Loss of material The relation between gravel and total traffic volume is shown in Figs 19-21 for the six sections. These show that the three sections that were completely reconstructed all exhibited similar rates of loss before and after reconstruction, with the loss by Sections 17 and 18 being sl .';_htly higher than for Section 19. The effect of maintenance grading on Sections 15 and 16 is clearly shown, the reduction-- in loss being due to surfacing material being brought back from the shoulders on to the carriageway. 11oa, 5 a90 80 70 60 50 40 30 20 10 7 Surm 150 425 148mnm 2.36 5 10 20 50 Particle size Fig.22 Change in grading of section 15 innCL 0. U90 80 70 60 50 40 30 20 10 U s 75Aum 150 425 148mnm 2.36 5 10 20 50 Particle size Fig.23 Change in grading of section 16 For the first 90,000 vehicle passes it was possible to compare the six sections directly. Up to this point Section 20 performed best having lost only 20 mm of material. This performance was closely followed by that of Sections 15 and 16 which each lost about 25 mm of material. The other three sections each lost approximately twice a's much material, Sections 17 and 18 losing around 70 mm and Section 19.losing around 45 mm. The performance of Sections 15 and 16 renamned good up to 140,000 vehicle passes by which time these sections had still only lost 35 mm of gravel. However, the performance of Section 20 rapidly deteriorated after 90,000 vehicles such that, by the time 140,000 vehicles had passed, the loss of material had more than trebled to 70 mm. However, measurements of rutting indicated that the verge-side rut in the Assela-bound lane hiad moved outside the normal width of the carriageway and it wasmost probable that this 'lost' material was actually deposited on the shoulder *just beyond the extreme point of the gravel loss measurements. Nevertheless, it would be normally extraordinary for a gravel road to carry this volume of traffic without even being graded. 481 T T=340 000 0 ~ T56 00 After reconstruction - Section 1 ~ Original construction TO0 T =3000 I I I~~~1 I T 0I 0 ~Afterreconstruction Section 18-. Scin 19 Original construction I I J t1 1 40P 100 90 80 70 70 60- 40 30 20 0 75Mmr 150 425 148mmn 2.36 5 10 20 so Particle size Fig.24 Change in grading of section 17 (Original construction) 00 90 70' -1.0 60 30 40 30 75p- 150 .425 143mm 2.36 5 10 20 50 Particle size Fig.25 Change in grading of section 17 after reconstruction 90 80 70 60 50 40 30 20 10 75Mum 150 425 148mm 2.36 5 10 20 50 Particle size Fig.26 Change in grading of section 18 (Original constructior 100 90 s0 ' 70 60 5 50 U40 ~; 30 20 10 U.- 75Mm 150 425 148mm 2.36 5- 10 20 50 Particle size Fig.27 Change in grading of section 18 after reconstruction 100 90 80 70 -T56,000 60 TO0 C T 0000 40- 30- 20 10 0 I I 7 5pmlSO0 425 148mm 2.36 5 10 20 s0 Particle size Fig.28 Change in grading of section 19 (Original construction CL 0. 90 80 70 60 50 40 30 20 10 0 75pm 150 425 143mm 2.36 5 10 20 so Particle size Fig.29 Change in grading of section 19 after reconstruction 100 90 80 c' 70 . 60 -50 c '. 40 (1 30 20 .10 0 75~m 150 425 148mmn 2.36 5 10 20 s0 Particle size Fig.30 Change in grading of section 20 9.4 Materials properties P~article-size distributions and compacted dry densities of the gravel sections were measured at intervals during the course of the experiment and are shown in Figs 22-30 and Table 6: For the untreated cinders in Sections 17 and 18, considerable changes of particle-size distribution were caused by trafficking during the first phase when the two sections were laid without compaction. When the two sections were reconstructed with compaction, the changes in 482 CL 10 T 32.000 T.185, 00 T -1 7.000 T
O 1 1 1 1 1 i 1 CL CL ----
---1 T z56. .. .000 0 .000 1 1 1 1 1 f L T 140,000 1 T T 30.000 1 1 1 1 1 1 1 1 1 T 32,000 T-85.000 7.000 T
O 1 1 1 1 1 Table 6. Density and moisture content of unpaved sections. Density (Mg/rn3) at months after construction Section 0 3 7 13 16 20 30 15 1.51 - 1.46** 1.46 - - 1.54 16 1.49 1 .54** 1.47 - -1.47 17 - 1.48** 1.56** 1.28* 1.45 1.55 - 18 - 1.46** 1.52** 1.24* 1.40 1.52 - 19 1.45 - .43** 1.39** 1.48 1.49 1.54 20 1.91 -1.83** 1.89 - - 1.94 Subgrade at construction = 1.49 Sub-base at construction = 1.49 Moisture content,' per cent, at months after construction 0 3 7 13 16 20 30 15 15 -9 12 -- 11 16 12 - 2 6 - - 3 17 - 9 1 13* 10 7 2 18 - 12 2 16* 11 4 2 19 12 - 4 16* 8 6 5 20 11 -4 12 -- 4 Subgrade at construction =15 Sub-base at construction =14 *After reconstruction **Measurements made on rough surface with no correction made for the irregularity particle-size distribution were less, but trafficking did lead to similar marked increases in density as in the first phase. The properties of the materials in the other sections, which had higher 'fines' contents, changed little during the experiment, although Section 19, the Mojo cinder with fines added, did change more than the others. 9.5 Conclusions from gravel surfaced sections The results of the gravel surfaced experimental sections showed that:- (i) After 140,000 vehicle passes and no maintenance, the Sodere agglomerate in Section 20 had performed well. However, during this time, as trafficking increased, the level of surface roughness and gravel loss became higher because abrasion of the fine soil matrix led to large oversize material (> 37 mm) being exposed and eventually lost from the surface. Screening or crushing the material to remove or break down the large-sized stone would be expected to improve the quality. (ii) In the main part of the experiment to examine cinder gravels it was clear that by compai~ing the modified cinders in Sections 16 and 19 with the untreated cinders in Sections 17 and 18 much better performance was obtained by mechanical stabilisation. This was reflected by lower rates of gravel~loss and roughness and by the development of corrugations. It was notable that the naturally occurring cinder gravel which contained sufficient fines and was used in Section 15 performed as well as the mechanically stabilised materials. However, such wellgraded cinder gravels are rare and it was regarded as an atypical material. (iii) There appeared to be little difference in the performance of the untreated cinder gravels (Sections 17 and 18) when comparisons were made between the materials laid with and without compaction. This was because traffic induced compaction accompanied by the breakdown of weak~ cinder particles increased the density to the same level in both cases. The fines produced in this way, however, were non-plastic and 483 Table. 7. Deterioration of premix surfaced sections. Section Measurement ~Years after construction Seton. esrmn 0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 Cracking 0 0 4.8 9.1 - 10.4 11.0 - (m/m' 12 Rutting 5 3 2 4 3 - 4 - (mm) Deflection 0.91 0.97 0.94 1.12 1.16 1.45 1.78 1.71. (mm) Cracking 0 0 1.3 9.6 - 8.3' 12.7 (1.0) (1.4) (3.6) 13 Rutting 3 3 .4 5 5 - 5 Deflection 1.06 1.23 1.38 1.47 1.37 1.45 1.64 1.49 (1.36) (1.18) (1.29) (1.33) (1.46) Cracking 0 0 0 0.7 - 4.6 15.9 - 14 Rutting 1 1 2 2 2 - 4- Deflection 0.79 - 0.78 1.03 1.02 1.18 1.3 1.30 Axle loading ESA x 1000 (Nazret 30 80 117 183 239 296 360 440 direction) Note: Values in brackets are measurements made on the half of Section 13 that was sealed with surface dressing did not bind the material as-well as the plastic volcanic ash soil. It is recommended therefore that in order to obtain good performance from cinder gravels as gravel wearing courses they should be stabilised with volcanic ash. (iv) In the mechanically stabilised sections of the experiment, approximately 10 per cent of the ash soil was dumped and spread by a front-end loader on top of the previously spread layer of cinder gravel. Mixing was carried out by motor grader and it is believed that this method of processing could readily be used in normal construction work. 10 PERFORMANCE OF THE PAVED TEST SECTIONS 10.1 Monitoring The paved sections were monitored for a period of 7X2 years. During this time, 440,000 cumulative equivalent standard axles were carried in the heavier trafficked lane towards Nazret. The rate of deterioration was studied without maintenance being carried out, except for Section 13 where, after 28 months, the first half of the section was treated with a singlesurface dressing to seal cracks. This enabled a determination of the effectiveness of the seal as cracks continued to develop on the other part of the section. The performance of the sections was assessed by surface measurements of rutting cracking and deflection at approximately yearly intervals. In addition, particle size distribution,. density and moisture content of the base materials and strength, in terms of CBR, and moisture content of the subgrade were measured three times during the experiment. The main difference between those sections that were treated with a double surface dressing and those treated with a premix surfacing are discussed below. 10.2 Rutting and cracking For the surface dressed sections (Nos 1- 11), there was no increase in rutting during the monitoring period and the maximum value recorded was 4 mm. For the premix sections (Nos 12-14), the rut depths were similar, but there was some indication that slight increases were occurring 484 ---- 7 as trafficking increased (see.Table 7). Of section Nos 1-11, the only cracking that occurred was on Section 11 where a longitudinal crack developed in the sixth year close to the edge of the pavenent in part of the Nazret lane. As there were no other signs of failure, it was believed that this was caused by shrinkage of the plastic volcanic tuff in the road base. Cracking began to occur in Section 12 of the premix sections in the third year and this quickly increased to a level regarded as unacceptable (TRRL Overseas Unit 1981). Section 13 also started to crack in the third year and the level of this had increased so much by the following year that the first half of the section was treated with a single coat of surface dressing, The effect of the surface dressing initially sealed the cracks but they were becoming extensive again three years later. The onset of cracking was .much later in Section 16 but, by the sixth year, these had also reached an unacceptable level., Examination of the premix surfacing in cracked areas showed clearly that the cracking had started from the top of the layer and had progressed downwards. 10.3 Deflection The'transient deflections of the paved sections are shown in Figs 31 and 32 -and those for the premix sections are also shown in Table 7. For the surfaced dressed sections, deflection measurements, which are recorded as the means of all values within a section at each survey, either reduced or stayed constant with increasing traffic during the first 6Y2 years of the experiment. A general increase occurred in the last series of measurements but, apart from Section 11, no other signs of surface deterioration .E 0 .t1.3 1.2 1.1 0.9 0.8 0.7 0.6 0.5 30 80 117 -183 239 296 360 440 Axle loading E.S.A x 1000 )Nazret Lane) Fig.31 Transient deflections Of surfaced dressed sections (Nos. 1-11) 1.8 1.7 1.6 IE .2* U 2 i1.5 ~- 1.4 1.3 1.2 1.1 1.0 0.9 .0.8 0.7 P irst half of section 13 sealed at 28 months / - I /~~~~~~~~~~~~~~~~~~~ 13 (first half) I , -. / Y / '14 Time (years) V.0 ~ -- ~35 . 556. . 0.5 1.5 2.5 I I 30 .80 117 183 239 296 360 Axle loading E.S.A x 1000 (Nazret Lane) 440 Fig.32 Transient deflections of premnix surfaced sections were evident. The behaviour of the premix sections was quite different, with Section 13 showing increases from early in the life of the experiment, and Sections 12 and 14 showing progressive increases from the third year onwards. The increases in deflection correlated well with increases in cracking. When surface dressing treatment was applied to part of Section 13 to seal cracks, this coincided with a reduction in the pavement deflection for two years. This was followed by further increases as cracks began to appear again. 10.4 Materials properties The measurements of materials properties in the base, sub-base and subgrade made at intervals during the experiment showed that the only marked changes that occurred were in the moisture contents and strengths of the pavement layers in the premix sections. However, it was not until the sixth year that increases in moisture content and reductions in strength were detected which was four years after cracks first appeared in the surfacing. This supports the observation that cracks in the surfacing started from the top of the layer and took some ,yearsbefore.they penetrated the whole layer. In situ CBR values of the different base materials, interpreted from the use of the Clegg Impact Hammer, ranged from 68 per cent to 140 per cent, with the highest values for the control section of the crushed stone base (Section 2) and the Sodere agglomerate (Section 4). Subgrade strengths for the volcanic ash soil varied from 13 per cent to 38 per cent at the time 485 Time (years) 0.5v~ 1.5 . 2.5 3.5 4.5 5.5 6.5 7. I I S 5.5 6.5 7.5 1 1 1 Table 8. In situ densities and moisture contents of the base of paved sections At construction After 1. year After 2Y2 years Section No. Density Moisture Density Moisture Density Moisture Mg/n3 ? content % Ng/nr' content % Mg/rn3 content % 1 1.53 14 1.43 13 1.47 11 2 1.99 7 1.91 4 2.05 4 3----- 4 1.90 9 1.71 6 1.89 4 5 1.49 9 1.32 7 1.48 5 6 1.43 10 1.35 6 1.50 5 7 1.41 10 1.46 2 1.46 1 8 1.59 15 1.45 6 1.47 5 9 1.50 16 1.60 8 1.57 7 10 1.34 15 1.34 4 - 2 11 1.'39 21 1.34 11 1.55 11 12. 1.45 12 1.58 ~ 5 1.38 6 13 1.45 14 1.36 9 1.36 9 14 1.48 16 1.47 14 1.44 11 5 m 1 a c1 .t0. 100 90 80 70- 60 50- 40- 30- 20 10 0 t~~1~ 75Dpm 150 425 148mmn 2.36 5 10 20 Particle size 50 Fig.33 Change in grading of section 6 (Bekojo cinder + fines) of construction. By the sixth year, the strengths of Sections 12 and 13 were 11 pek cent. No appreciable changes occurred in the particle size distributions or the densities of the base materials. As examples, the particle size distributions of Sections 6 and 9 measured at four different tines are shown in Figs 33 and 34 and densities and moisture contents are shown in Table 8. 10.5 Conclusions from the paved test sections (i) The satisfactory performance of the surface dressed sections showed that cinder gravels, whether untreated or mechanically stabilised with volcanic ash soil, were suitable for use in road bases up to a 75jum 150 425 148mmn 2.36 5 10 Particle size Fig.34 Change in grading of section 9 .(Crushed Mojo cinder 4- fines) 20 s0 level of 440,000 esa. The use of quarry fines (less than 5 mm maximum size) vibrated into the surface of the compacted cinder gravel also provided a satisfactory base material. (ii) The mechanically stabilised cinder gravels and the km 130 material which was naturally well graded were easier to compact than untreated cinders. ..(iii.) .The dry *bo'und macadam, the Sodere agglomerate and the volcanic tuff also performed satisfactorily during the experiment although the volcanic tuff did show some cr~acking. (iv) The roadmix surfacing was unsatisfactory. It is believed that the material was too porous and that either changes in the properties of the bituminous binder, insufficient binder or inadequate mixing was the main reason that caused 486 ---- 7 the surface to crack. It is possible that the application of a surface dressing to the roadmix immediately after construction could have prevented the cracking. However, with the considerably better performance of the surface dressed Sections 1-11, the use of roadmixed asphalt is clearly uneconomic with or without a further seal. 11 RECOMMENDATIONS i)Cinder gravels typical of those used in the full-scale road trials and with particle size distributions similar to those shown in Fig 1 can be used for road bases in lightly trafficked surface dressed * roads. Suich gravels have now been tested * for carrying a traffic loading of 440,000 esa, which is close to the design for 500,000 esa included in Road Note 31. Cii) Other materials used in the road base of the surface dressed sections performed satisfactorily. These included dry bound macadam, which is a construction process without the use of water, Sodere agglomerate and a volcanic tuff. (iii) For gravel surfaced roads, mechanically stabilised cinder gravels perform much better than as-dug cinders which lack plastic binder even after the breakdown of cinder particles by trafficking. Recommended gradings for materials which are more resistant to corrugations are shown in Fig 15. (iv) Mechanically stabilised cinder gravels using 10 per cent of locally available volcanic ash soil are easier to compact and process during construction than as-dug cinders. For as-dug cinders, it is necessary to start compaction with * a pneumatic tyred roller before using either a smooth wheeled or vibrating roller. There was little breakdown of cinder particles during compaction with the type of plant used. (v) Roadmix is not a satisfactory method of providing a bituminous surface for cinder gravels. 12 ACKNOWLEDGEMENTS This paper is published by permission of the Director of Transport and Road Research Laboratory, United Kingdom and the General Manager of the Ethiopian Transport Construction Authority. The work was carried out as part of the joint research programme of the two organisations and the authors wish to thank members of the team who participated in the project. REFERENCES BSI, 1971. Specification for singlesized road-stone and chippings. BS63: 1971 (Part 2). London: British Standards Institution. BS1, 1975. Methods of test for soils for civil engineering purposes. BS1377: 1975. London: British Standards Institution. Clegg, B. 1976. An impact testing device for in situ base course evaluation. In: ARRB. Proc. of. the 8th Conference of ARRB, Perth, 1976. Vermont South: Australian Road Research Board. Heath, W. & R.Robinson 1980. Review of published research into the formation of corrugations on unpaved roads. Supplementary Report 610. Crowthorne: Transport and Road Research Laboratory. Hendrickson, L.G. & J.W.Lund 1970. Construction specifications for volcanic cinders used as road surfacing aggregate. Highway Research Record No 307. Washington DC: Highway Research Board, National Research Council. Kleyn, E.G. & P.F.Savage 1982. The application of the pavement DCP to determine the bearing properties and performance of road pavements. In: Nordal, R.S. and R. Saetersdal. International Symposium on Bearing Capacity of Roads and Airfields, Trondheim, 23-25 June 1982, Proceedings Vol 1. Trondheim: Norwegian Institute of Technology, 238-246. Mesfin Wolde-Marian 1970. An atlas of Ethiopia. Addis Abeba: Haile Selassie I University, Revised edition. Newill, D.& Kassaye Aklilu 1980. The --location and engineering properties of volcanic cinder gravels in Ethiopia. In: Anon. 7th Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Accra, Ghana, 1-7 June 1980. Rotterdam: AA Balkema, 21-32. Smith, H.R. & C.R.Jones 1980. Measurement of pavement deflections in tropical and sub-tropical climates. TRRL Laboratory Report 935. Crowthorne: Transport and Road Research Laboratory. Transport and Road Research Laboratory 1977. A guide to the .structural design of bituminous-surfaced roads in tropical sub-tropical countries. Road Note 31. London: HMSO, Third edition. TRRL Overseas Unit 1981. Maintenance management for district engineers. Overseas Road Note 1. Crowthorne: 487 Transport and Road Research Laboratory. TRRL Overseas Unit 1985. Maintenance techniques for district engineers.. Overseas Road Note 2. Crowthorne: Transport and Road Research Laboratory, Second edition. Crown Copyright. Any- views expressed in this paper are not necessarily those of the Department of Transport. Extracts from the text may be reproduced, except for commercial purposes, provided the source is acknowledged. 488