Department For. E I I~~~ I L .~~ International D F I13~~~Development TITLE: by: Use of Soft Limestone for Road-Base Construction in Belize M E Woodbridge Transport Research Laboratory Crowthorne Berkshire RG45 6AU United Kingdom PA3450199 PA3450/99 WOODBRIDGE, M E (1999). Use of soft limestone for road-base construction in Belize. Seventh International Conference on Low-Volume Roads, Baton Rouge, Louisiana, USA. 23 -26 May 1999 TRANSPORTATION RESEARCH RECORD 1652 8 Use of Soft Limestone for Road-Base Construction in Belize M. E. WOODBRIDGE The results of a highway experiment, constructed in May 1978 in north- ern Belize, designed to investigate the suitability of locally occurring cal- careous materials, known as manls, for road bases arc discussed. The manls comprise high-purity carbonate materials containing mainly silt- sized particles and fall outside the grading, plasticity, and strength spec- ifications normally required for road bases. Three manls, each with slightly different characteristics, were substituted as road base for crushed stone. One of the marls Was also stabilized with ordinary port- land cement. Detailed monitoring was then undertaken periodically to determine their performance. The road pavement was constructed on an embankment to ensure good drainage. A good quality surface dressing seal has been maintained. After 19 years of traffic, measured at 1.3 mil- lion equivalent standard axles, the marl road bases have performed at least as well as the crushed stone. The cement-stabilized marl road base performed exceptionally well. Stabilization would enable the use of more plastic marls. Belize is a small country situated on the Caribbean coast of Central America. Nearly 23 000 km2in area, it is about 1/10th the size of the United Kingdom and has a common border with Mexico and Guatemala. It became fully independent in 198 1, although it has been internally self-governing since 1964. The country has four main highways (Figure 1) totaling 600 kmn and now built to bituminous standard. Belize has a subtropical climate with a pronounced wet season. Data supplied by Belize Sugar Industries (BSI) indicate that average annual rainfall for the period 1974l-1989 was 1364 mm, of which 72 percent (983 mm) fell between June and October. Monthly day --maxima temperatures vary from 300C to 35'C and night minima tem- peratures vary from 15C to 250C. Relative humidity usually exceeds 80 percent. In December 1973 Sir William HalcroW and Partners (the con- sultant) were appointed under United Kingdom technical aid to carry out a feasibility study for improvement of the Northern High- way. They initially considered using marl for road base because of the good performance of existing marl pavements in Belize and Mexico, but marl was eventually rejected because of the low wet strength, poor grading, and relatively high plasticity; crushed stone was used as road base instead. The Transport Research Laboratory (TRL) was invited to carry out further investigations. A full laboratory investigation was carried out and a field trial was constructed in May 1978, using three marls sub- stituted for the crushed stone road base used in the main contract, including one marl stabilized with cement. The performance of the four trial sections was compared with a control section of crushed stone. Monitoring visits were made in March 1979, October 1980, March 1984, October 1989, November 1992, and September 1997. This report describes the following: * Selection, engineering classification, and composition of the marls; * Design parameters and construction of the road trial; * Results of the monitoring visits; and * Engineering criteria for future use of the marls. SELECTION, ENGINEERING CLASSIFICATION, AND COMPOSITION Selection The marl occurs ubiquitously in northern Belize and the Yucatan beneath a thin layer of clay. It is underlain by limestone at depths varying from 4 to 9 m. The consultant identified several marl borrow areas for fill material from which TRL selected three for road bases: Tower Hill (or Texaco Pit), San Victor (or Buena Vista North), and Santa Cruz. Engineering Classification of the Selected Marls Compaction Laboratory compaction tests were carried out in 1978 on samples taken from stockpiles. The results are presented in Table 1. The unsoaked California bearing-ratio (CBR) values of the marls are very high and comparable to the crushed stone but their soaked CBR values are much lower. This characteristic caused them to be rejected for road base. Grading Figure 2 presents the results of gradings carried out on the stock- piled marls. All were outside the recommended grading envelope for mechanically stable natural gravel. Gradings determined on marl samples taken after comnpaction (Figure 3) were even finer grained than the stockpiled marl samples and therefore further out- side the recommended grading envelope. This reduction in grain size was undoubtedly caused during construction of the trial sec- tions. Normally such a feature would be associated with significant deficiencies in performance. Plasticity Transport Research Laboratory, Old Wokingham Road, Crowthorne, United Kingdom. Atterberg limits were determined on the fraction passing the British standard (BS) 425-jgtm sieve (1) of the stockpiled marls and the results . 181 Location of roadbase trial + (Jo~~~~~~~~~ AmbergrisCay --- A 5' wI I I I WI a I 0 p a1 40 PUNTA GORDA Islands o 10 . 20 FIGURE 1 Belize: Marl road-base trial, general location. vniles Woodbridge13 TABLE 1 Laboratory Compaction Characteristics of the Selected Marls LOCATION SECTION 1 SECTION 2 SECTION 3 SECTIONS 4 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~~~~~~~~~~~~(&5) MATERIAL CRUSHED MARL, MARL, MARL, STONE, TOWER HILL SAN VICTOR SANTA CRUZ SAN ANTNIO ____1 __ ____________________ 1978 1984 1978 1984 1978 1984 1978 1984 BS (Heavy) Com ctn Max. Dry Density Mg/rn 3 2.24 2.24 1.97 2.00 1.67 1.65 1.70 1.70 (MDD)Optimum moisture 4.0 5.5 9.5 9.2 16.0 16.0 16.0 15.6 Content % (OMC) at OMC 140 130 180 140 150 130 160 Soaked (4 days) 140 50 40 60 30 40 50 BS (Light) Compaction Max. Dry Density, Mg/rn 3 2.15 2.17 1.86 1.89 1.51 1.61 1.62 1.63 Optimum Moisture Content % (OMC) 9.0 8.0 11.0 12.6 21.0 20.3 18.0 16.0 At OMC 120 35 18 45 14 25 15 Soaked (4 days) 80 16 1 1 16 5 16 8 are presented in Table 2. The marls had a high proportion of fines but did not show high plasticity. According to TRL Road Note 31 (2), the plasticity index of the material passing the BS 425-lgtm sieve should not exceed 6 for road base and the liquid limit should not exceed 25. Under these criteria, the marls were of marginal quality for road-base materials. [In the fourth edition of TRL Road Note 31 (3), the quantity of plastic material, expressed as the plasticity modulus or the product of the plasticity index and percentage passing the BS 425-gin sieve, is specified, with a limit of 90: the marls have much higher values.] Swell was measured during determination of soaked CBR: val- ues of 0.1 percent were recorded, indicating that the fines were nonswelling. Chemical and Petrologic Composition The chemical and mineral constituents of the marls were determined and are indicated in Table 3. The results confirmed that the marls consist almost entirely of calcium carbonate and that the crushed stone is dolomite. In fact, the makerials are not marls, which are cal- careous clays or shales, but limestones. The term marl is retalned, however, because that is how they are known locally. Although the marls have a high fines content, the high carbonate content has resulted in low plasticity and a softness of the constituent particles. The chemical composition has also profoundly affected the subsequent performance of the marls. In terms of grain size, they are akin to silt or loess, but, whereas loess typically comprises hard par- ticles of rounded silica between 0. 15 and 0.075 mmn and is liable to have a low shear strength because of the poor grading and particle shape, the marl comprises soft particles that break down and lock together on compression or compaction, giving high strength en masse. This is the feature that has led to the good performance of the marl, as discussed below. Thin sections of the marls were examined (Figures 4 and 5) and showed fine-grained calcareous particles agglomerated around a harder particle or a shell fragment. The larger particles were porous, allowing the resin with which the slides were made to permeate them quite easily. The microporosity and softness are important features of the marls. When compacted, themarl particles meld together to form a strong and compact mass. Water can, however, permeate them quite easily and, although the carbonate minerals do not have the property of accommodating water within them, as clay minerals do, the water instead forms a surrounding film, which reduces the mass strength. Microfossils occur, chiefly foraminifera, which are of marine origin. The marls are typical of the oozes, which have a widespread occur- rence in the tropical oceans of today. They are also physicochemically similar to the Middle and Upper Chalk of western Europe (4). Chalk is obviously more indurated but, when broken down into its elemen- tal silt- and clay-sized particles, has liquid and plastic limits of 27 and 21, respectively, and therefore a plasticity index of 6. It also has a notoriously low wet strength in its disaggregated state. ROAD DESIGN AND CONSTRUCTION Design The consultant assembled data from a variety of sources and esti- mated a cumulative figure of 0.5 million equivalent standard axles (esa) for the heaviest trafficked lane, south to Orange Walk town, for the 10-year design life. Low areas existed along the road alignment of Section A, which were seasonally inundated. To raise the pavement above flood level, construction of an embankment was proposed, elevated from 400 to 800 mm above ground level. The proposed embankment fill was the marl obtained from appropriately spaced borrow pits along the road alignment. Soaked CBR values of this marl at BS (heavy) comn- paction were at least 10 percent, indicating its suitability for fill, according to the third edition of Road Note 31 (2). .....- .1 .. ...,. 183 184 ~~~~~~~~~~~~~~~~~~~TRANSPORTATION RESEARCH RECORD 1652 C,,'C,, rE c)LO c:' a C' C:0 M ) 9' W I I -Y--1---I--- I- --T I - ' I 1 C) 0 C),-cmN.C CD U) CD c o1 u- 0 N . C) c, - ~~~Imperial c I I BRITISH STANDARD SIEVE SIZES PARTICLE SIZE -mm t"-C.) C' C1 , I- I~ ri ( quivalent sieve sizes I I SAND GRAVEL COBBLES FIE EDIUM 1COARSE I FINE MEDIUM COARSE 2 6 PARTICLE SIZE -mm 20 60 200 Note: 1: Section One, San Antonio crushed stone 2: SectionTwo. Tower Hill or Texato Pit marl 3: Section Three, San Victor marl 4 and 5: Sections Four and Five, Santa Cruz or Buena Vista North mart FIGURE 2 Belize: Marl road-base trial, grading of stockpiled mans before construction (May 1978). For subbase, the marl from the Santa Cruz borrow pit satisfied the minimum design CBR value of 25 percent. For road base, the materials were required to have a minimum soaked CBR value of 80 percent at BS (light) compaction. The marl was thus precluded for use as road base. The only material available at that time that satisfied this specification was the crushed stone from the San Antonio quarry. The target dry densities relative to the maximum dry density, BS (heavy) compaction, were 90 percent for the embankment fill and 95 percent for both the subbase and thc road base. The proposed surfacing consisted of a prime coat blinded with stone dust followed by an "inverted" double surface dressing-that is, a larger-sizedistone layer laid over a smaller-sized stone layer, applied over the central 6.7 m of carriageway. The road shoulders, each 1.83 m wide, were to be left unsealed. The surface dressing con- sisted of I13-nun chippings, followed by 20-mm chippings, both obtained from the San Antonio quarry. This stone had aggregate impact values ranging from 19 to 26 percent but was not of ideal qual- ity for surfacing aggregate because it was susceptible to polishing. However, there was no alternative source. Construction The TRL experimental sections were constructed in May 1978 dur- ing the main contract. Embankment and subbase construction was 01C', E CD  E F. 0 C') C')10 N I - I - 100 90 -180 C')iul 0 CON r~iI I - -4 0C' 0- 70 -gm 60 0m2 -g 40 > z 30 I. -j-120 10 0 ~~2 ' ' ' ~1/1 I4 .06 .2 .6 184 1 i PARTICLE SIZE -mm o 0 o C> 0C. BIRFISH STANDARD SIEVE SIZES E E ~ s'! 8 a 8 cl mn -- CJCIt 100 I 111 __lC -Section 1 9 90 -(crushed stone) 9 80 80 70 70 60 60 50 (from stockpile) 7 0 TTTM. ~~~~~~~~~~~~~~~~~~~~~~~40 30 20 ~~~~~~~~~~~~~~~~~~~~~~~~~~~20 10 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~10 0 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~0 PARTICLE SIZE -mm BRMlSH STANDARD SIEVE SIZES 90 mart)- -~~~~~~~~0C4 ? -0')' 60 ** ate.cntrcton6 -Scto beoecntuto 10 11:~~~~~~~~~~~~~~~~~M1 80 0 FIGURE 3 Belize: Marl road-base trial, grading of marls before (1978) and after (1984 and.1989) construction. r2 00 186 ~~~~~~~~~~~~~~~~~~~TRANSPORTATION RESEARCH RECORD 1652 TABLE 2 Atterberg Limits of Stockpiled Marts SECTION 1 SECTION 2 SECTION 3 SECTION 4 Crushed Stone Marl, Tower Marl, San Marl, Sta. Hill Victor Crurz Mass percentage passing 425 17 75 55 62 micrometer sieve_______________ Liquid limit n/p 24 29 29 Plastic limit n/p 15 23 23 Plasticity Index -9 6 6 Plasticity modulus -675 33037 carried out by the contractor. The marls used for road base were then laid under TRL supervision. Five sections, each 106.7 m (350 ft) long, were laid. Each included a 15.2-rn (50-ft) transition zone between sections, leaving 91.4 m (300 ft) for monitoring purposes. Section 1 was a control where the road base consisted of the crushed stone used in the main contract; Section 2 was marl from the Texaco, Pit; Section 3 was marl from the San Victor pit; Section 4 was marl from the Santa Cruz pit; and Section 5 was the Santa Cruz marl stabilized with cement. The experimental site was located in a low swampy area. From 600 to 750 mm of marl fill was laid and compacted followed by 100-mm- thick subbase of Santa Crutz marl. Construction of the marl road bases was then completed -under TRL supervision. Section 5, the cement- stabilized road base, .was laid separately. After the marl had been dumped and spread, Mexican portland cement bags were placed at a predetermined spacing equivalent to 5 percent by weight. The cement was manually spread and mixed into the marl with a grader. Water was then added before further mixing and spreading and the material was then compacted with smooth-wheeled rollers for 1.5 h. The total time taken for the cement-stabilizing operation was 3 h. The compacted marl road-base thickness was later checked and found to be within ±5 mmn of the design thickness of 150 mm, except Section 5, whose mean thickness was 180 mmn. After the road bases had been constructed, a prime coat of MC 30 was applied at 0.61 Tim 2on Sections I to 4. On the cement- stabilized section, a higher viscosity prime, MC 3000, was used to facilitate curing. This prime was slow to harden and also slig htly damaged by unauthorized traffic, so it was decided to apply the first layer of binder and surface dressing chippings before it had hardened. TABLE 3 Chemical Composition of Crushed Stone and Marns Description San Antonio Tower Hill San Victor Santa Cruz Dolomite or Texaco Pit Marl Marl _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ M arl _ _ _ _ _ _ Section 1 2 3 4& 5 Sio 2 0.6 2.7 2.2 2.3 A1,0, 0.2 0.7 0.4 0.5 Fe2O, 0.05 0.15s 0.1 0.2 CaO 35.4 52.3 53.6 53.0 MgO 17.4 1.2 0.5 0.7 Mn203 0.01 0.02 0.01 0.01 P205 0.01 0.01 0.01 0.01 TiO2 0.01 0.02 0.02 0.02 K20 0.02 0.03 0.02 0.04 LO01 45.9 42.3 42.9 42.7 Total 99.6 99.4 99.8 99.5 CaCO 3equiv 3 63.2 93.4 95.7 94.6 MgC0 3equiv 3 36.5 2.5 1.0 1.5 Notes:Components determined by X Ray fluorescence on a fused bead sample, undertaken in duplicate 2 Loss on Ignition at 1000'C CaCO 3equiv is CaO x 1.785 MgCO, equiv is MgO x 2.09 186 Woodbridge17 U FIGURE 4 San Victor marl under crossed polar light, showing nucleation growth of one particle but nonnucleation in others; dark areas are voids. On the other sections, the surface dressing was applied within 2 days of priming. A layer of 13-mm chippings was applied on MC 3000 (0.95 UM 2). A second layer of 20-mmn chippings was applied during the main construction period 3 months later. After construction of the field trial, two boreholes were sunk and lined with perforated plastic tube in order to determine the fluctua- tion of the water table at the site. Observations carried out during 1978 and 1979 showed that the water table was about 1.5 m below ground level in the dry season. In the wet season, however, water -overflowed the top of one borehole and was near the top in the other. Standing water was also present in the side drains. MONITORING OF EXPERIMENTAL SECTIONS Traffic Surveys It was established that 850 vehicles per day used the road in 1992 compared with about 600 per day in 1978, which represents an annual growth rate of 2.5 percent over this period. The traffic gener- ated by the sugar cane harvest is very significant because it accounts for a large proportion of. the commercial traffic on the road. Traffic volumes are increased by approximately 200 vehicles per day during the cane-harvesting season, although this figure has obviously fluc- FIGURE 5 Same as Figure 4 but with ultraviolet light (lighter areas are voids), showing microporosity of the particles. tuated from year to year according to economic anji climatic factors. Almost all sugar grown in Belize is cultivated between Orange Walk and Corozal and, since 1986, most has been transported southward to the Tower Hill factory near Orange Walk town. The annual esa, presented in Table 4, have been calculated from data supplied by BSI and are verified by the TRL surveys in 1980, 1984, and 1992 (boldface type). A loaded sugar cane truck typically weighs 12 Mg (12 tonnes), 8.5 Mg (8.5 tonnes) of which is sugar cane. Approximately three-quarters of the load is carried by the rear axle; therefore, each loaded truck carries about 1 esa. From BSI records, the annual esa contribution for both sides of the road trial caused by sugar cane movements to the Tower Hill and Libertad sugar refineries has been calculated. In 1986 sugar production was centralized at the Orange Walk refinery and the Libertad refinery was taken out of service, with obvious effects on the axle loads. The values presented in Table 3 for the intervening years between the surveys were interpolated by using the 2.5 percent average growth rate given previously. Performance of Marl Road Bases Field Monitoring The density attained at construction and changes in moisture con- tent are the main factors influencing the strength of pavement mate- rials. Water infiltration through the surface or from below can result in a progressive loss of strength of the pavement materials and con- sequent deterioration of the- road. Therefore, one of the principal aims of road materials designizs to minimize the ingress of water. At construction, the dry density varied ±3.5 percent from the average value. The variation in moisture content was ±30 percent from the average value. During each monitoring visit, a number of test holes were dug in each trial section to determine the variation in moisture content and strength across the width of the road. There were five sampling positions: both road shoulders, verge side wheel tracks, and the center line. At each position a dynamic cone penetrometer profile or a Clegg impact test or both were car- ried out. Then a sand replacement density and moisture content were determined on the road base. The hole was continued down through the subbase and embankment fill by using a screw or post- hole auger and was terminated in or just above the natural ground. Samples were removed at intervals and their moisture content was determined. The main trends in the moisture content, dry density, and strength of the marl road bases are summarized below: 1. There has been a slight increase in moisture content in the road bases with time but average strengths are still high. The crushed stone section is the strongest, and the marl is less strong (CBR, 60 to 100 percent), especially the unsealed road shoulders. 2. The moisture content of the marl subbase has increased slightly compared with the construction moisture, but the strength is still high (CBR, 80 to > 100 percent).. 3. The thickness of marl fill laid under the road trial varied from about 600 to 700 mm. The construction moisture content of the fill was approximately 13 percent and it has wetted up only slightly since construction, more in Sections 1 and 2 than in Sections 3 and 4. 4. In all sections, but particularly Sections 3 and 4, the very high CBR values of the fill between 300 and 450 mm occur because of 187 1 -a 188 ~~~~~~~~~~~~~~~~~~~TRANSPORTATION RESEARCH RECORD 1652 TABLE 4 Annual Axle Loads for Trial Sections Year Orange Walk lane, esa Corozal lane, esa cane traffic non-cane cane traffic non-cane traffic traffic 1979 6,900 17,000 15,800 8,000 1980 14,300 7,000? 14,600 8,000 1981 12,700 19,000 14,000 8,000 1982 11,500 20,000 16,000 8,000 1983 13,000 21,000 15,,600 8,000 1984 12,700 22,000 14,000 8,500 1985 11,300 23,000 13,900 9,000 1986 47,300 24,000 500 10,000 1987 42,900 25,000 500 11,000 1988 43,000 26,000 500 12,000 1989 47,100 27,000 2000 13,000 1990 51,400 28,000 4300 14,000 1991 51,200 29,000 4700 15,000 1992 52,400 29,000 2600 15,500 1993 51,300 30,000 4600 15,500 1994 54,500 31,000 3500 16,000 Totals 523,500 378,000 127,100 192,000 Grand 901,500 319,100 Totals the presence of a stony layer. It appears that the stonier marl was fortuitously extracted from a local borrow pit during embankment construction. The presence of the stoney layer has significantly improved the overall strength of the pavement, measured in terms of the structural number, as discussed below. Structural Number The concept of structural number was developed as a result of the AASHO road test (5) to assign. indices to road pavements -derived from the strengths and thicknesses of their constituent layers. Structural number =aidi where ai = strength coefficient of pavement layer i, and di= thickness of pavement layer i (mm). The strength coefficients of the road base and subbase were worked out for various types of unbound and bound materials and are sum- marized by Paterson (6). In the AASHO0 road test, the strength of the subgrade was CBR 3 percent, so that the merits of combinations of different pavement materials could be assessed. However, to take account of subgrades with different strengths, the concept of modified structural number was developed by Hodges et al. (7). Further adjust- menit to allow for the decreasing influence of subgrade strength with depth was made by Rolt (8). From the results of this trial, the average CBR values of the pavement layers have been estimated and are pre- sented in Table 5. Average modified structural numbers have been calculated from the visits with the corresponding strength coefficients derived from Paterson (6). The fourth edition of TRL's Road Note 31 (3) contains a struc- tural catalogue where the subgrade strength for design is assigned to one of six classes, The subgrade (i.e., fill) CBR of all the sections is >30, which places them in category S6 (the strongest subgrade) of Chart 1-that is, the chart for granular road bases with surface dress- ings. The modified structural number for each section is then related to a prediction of the traffic that section should withstand (i.e., the design traffic). According to these predictions, all the sections have yet to reach their design traffic, mainly because of the very large contribution made by the subgrade marl. Deformation The repetition of cyclic stresses of varying intensity over a period of time results in the development of permanent strains whose magni- tude depends on the original strength of the pavement and subgrade and the changes that have occurred in these materials with time, such as densification. The permanent strains are manifested by rutting in the parts of the road (i.e., the wheel tracks) subject to the cyclic load- 188 Woodbridge18 TABLES Pavement Layer CBR and Structural Number Section Roadbase Sub-base Subgrade MSN' MSN 4 Traffic CBR, % CBR, % CBR, % (of section) (of cat) Class 2 1 > 100 70 >30 3.15 > 3.05 T5 2 70 70 >30 2.93 > 2.92 T4 3 60 > 100 >30 2.93 > 2.92 T4 4 70 > 100 >30 2.93 > 2.92 T4 5 CB2 3 > 100 >30 3.32 > 3.18 T6 1. Modified Structural Number is: MSN = 0.0395 (sum a1..d,) + SGC SOC is Subgrade Contribution is: SGC = 3.51.1og CBR -0.85 (log CBR) 2-1.43 2.T4 = 1.5 to 3.0 million equivalent standard axles T5 = 3.0 to 6.0 million equivalent standard axles T6 = 6.0 to 10 million equivalent standard axles 3. CB2 is cement stabilised roadbase, UCS = 3 to 6 MN/rn 2 4. Overseas Road Note 31 (1993), Chart 1 ings. The magnitude of the elastic deflection provides a measure of the performance and residual life of the pavement according to generally recognized limits. Rut depth and deflection measurements were determined at approximately 8-rn intervals in the wheel tracks of both traffic lanes during each monitoring visit: any changes in the measurements can thus be related to increasing traffic or to changes in the properties of the pavement or subgrade. Rut depth measurements were carried out with a 2-rn straightedge laid across the verge side and offside wheel paths in both lanes. Results for the last two visits are summarized in Table 6. Although there has been a recent slight increase in deformation, it is not appre- ciable and is typical of the kind of variation found in surface-dressed roads at the time of construction. Usually, rut depths ranging from 1 5 to 25 mm are required before the pavement is considered to be in a critical condition. These small values of rut depth are all the more remarkable because the density levels achieved in the road base at construction were about 95 percent relative to BS (heavy) compaction, which normally is considered quite low. Clearly, however, there has been little secondary compaction under traffic. Deflection was measured by a standard TRL method with the rear axle of a truck loaded to 62.3 kN.- Measurements were made with a Benkelman beam, a device consisting of a slender metal beam that measures the flexing of the road pavement as a loaded truck passes. Deflection measurements are vefy small; values of 1 to 1.5 mm are high and can indicate a critical condition (6). Deflection measurements are summarized in Table 7. They are all remarkably small. However, all sections have shown slight increases with time, with Section 1 (the control) showing the highest and Sec- tion 5 (the cement-stabilized marl) showing the lowest. The crushed- stone road-base deflection values are typical of this material. All the marl road-base values, however, are typical of cement-stabilized materials or even concrete. The measurements of deflection and rut depth were accompanied by visual observation of the general condition of the trial road sec- tions. In 1997 all the trial sections were in excellent condition. No maintenance of the pavement had been necessary other than the resur- facing, although on the shoulders marl has been replaced periodically to compensate for erosion losses. TABLE 6 Rut Depth Data Section Uale M~~range Wvalk lane Cooail ______________________ Vergeside Offside Offside Vergeside 1 Nov 1992 6 (3-8) 4 (3-8) 5 (4-8) 6 (3-10) ____________ Sep 1997 12 (5-15) 6 (5-8) 5 (4-8) 8 (5-10) 2 Nov 1992 7 (4-11) 5 (1-8) 4 (3-6) 1 (0-4) ____________Sep 1997 12 (5-15) 7 (5-10) 5 (5-6) 5 (5-7) 3 Nov 1992 5 (3-8) 5 (4-7) 4 (3-5) 5 (3-7) ____________ Sep 1997 5 (3-7) 5 (2-7) 5 (3-6) 5 (3-7) 4 Nov 1992 6 (4-8) 5 (3-6) 5 (4-7) 6 (4-7) ____________Sep 1997 7 (4-9) 5 (3-7) 6 (4-8) 6 (4-10) _____________ Sep 1997 110 (5-15) [7 (4-7) J4 (3-8) (35 Notesl. Averages are single figures, ranges are in brackets. 2. Each average is calculated from nine measurements taken at the crosses in Figure 7. -,x. ... ,.-... .. . .. ..  ?  -. ,.. ... .- '-... 41 Notes 189 190 ~~~~~~~~~~~~~~~~~~~TRANSPORTATION RESEARCFI RECORD 1652 *TABLE 7 Deflection Measurements SECTION YEAR ORANGE WALK LANE COROZAL LANE Vergeside Inside Inside Vergeside Wheeltrack Wheeltrack Wheeltrack Wheeltrack mm mm X 10.2 mm X 102 mm X 10.2 X 10.2 MAR 79 39(29-48) 29(24-38) 27(23-36) 31(22-39) (Control) OCT 80 41(35-51) 36(30-46) 42(34-SO) 41(3.0-49) MAR 84 44(39-55) 53(43-64) 34(28-38) 48(43-53) OCT 89 62(54-86) 5 1(47-59) 45(37-54) 48(43-53) NOV 92 64(54-75) 63(51-80) 43(39-SO) 58(50-70) 2 MAR 79 25(24-31) 21(18-25) 17(12-30) 17(11-21) (Texaco Pit OCT 80 22(13-38) 24(18-33) 17(11-27) 17(14-24) Marl) MAR 84 28(17-36) 35(24-40) 17(13-22) 28(17-39) OCT 89 31(20-45) 29(18-34) 24(16-34) 27(20-38) NOV 92 51(40-61) 49(36-61) 23(19-27) 35(20-50) 3 MAR 79 13(11-17) 13(12-15) 13(11-18) 11(9-15) (San Victor OCT 80 12(9-18) 12( 7-18) 7(4-10) 9( 5-13) Marl) MAR 84 10(7-15) 21(18-25) 13(10-19) 16( 9-22) OCT 89 16(12-19) 16(14-20) 16(14-21) 23(19-30) NOV 92 21(15-28) 27(17-53) 22(16-30) 18(15-20) 4 MAR 79 12( 8-14) 12(7-14) 12(11-15) 9( 5-13) (Sta Cruz OCT 80 11( 9-1.3) 10(7-1 1) 9(7-12) 10( 7-12) Marl MAR 89 10( 8-11) 15(9-19) 13( 7-16) 8( 8-9) OCT 89 15(10-18) 12(8-15) 13( 9-16) 18(15-24) NOV 92 19(11-24) 18(15-21) 19(9-23) 15(12-17) S MAR 79 7( 6-11) 9(7-15) 9( 6-19) 9( 6-12) (Cement stab. OCT 80 9( 6-12) 11(8-14) 9(5-14) 9(3-13) Sta Cruz MAR 84 9( 6-12) 13(8-19) 9 (4-20) 14( 6-30) marl) OCT 89 13(10-20) 14(6-20) 17(12-25) 17(12-24) ____________ NOV 92 18(15-22) 16(10-23) , 23(16-53) 15(10-21) Note: Each single value (range in brackets) represents the average of nine duplicated readings. March is towards end of dry season, when the pavement is strongest: October is at end of wet season, when the pavement is weakest. Performance of Cement-Stabilized Marl Road Base The Santa Cruz marl. was selected for stabilization with 5 percent cement. In the laboratory two levels of static compaction were used for the stabilized marl to give 92 and 97 percent of the maximum dry density obtained in BS (heavy) compaction. These two levels were chosen to embrace the specified density level of field comn- paction (95 percent). Unconfined compressive strength (UCS) val- ues ranging from 2.5 to 5.5 MPa were obtained for the marl. after 28 days of curing. According to Road Note 31 (9) the stabilized material would be classified as CB 1 [i.e., within a UCS range of 3.0 to 6.0 MPa (1 MN/in 2= 1 MPa)]. The performance of the cement-stabilized marl road base has been excellent. A number of cracks developed in the early years but these have not increased. The low values of rut depth and deflection testify to the high strength of the road base. In 1992 samples were tested giving UCS values averaging 9 MPa for the fresh stabilized marl beneath the surface dressing and 5 MPa for the partially car- bonated marl from the road shoulders. Strength increases with time are usual in uncarbonated stabilized materials and would obviously enhance the durability of the pavement because, apart from the grad- ual strength gain, the material becomes more water resistant. This result indicates that it is possible to use a much wider range of marls if they are stabilized. Performance of Surface Dressing The condition of the surface dressing has been crucial to the perfor- mance of the road bases. The original surface dressing was in very good condition during the 1989 visit (it was then 1 1 years old), with very little cracking, bleeding, and loss of stone. It was sampled and the bitumen was recovered by the Rotovapor method (9). Pene- tration tests were carried out, with the following averaged results [Pen units (mm x 1 0-)]: Section 1, 32; Section 2, 31; Section 3, 25; Section 4, 26. Although the bitumen recovered was a composite of the prime and two surface dressings it had clearly undergone some hardening but still retained considerable flexibility. MC 3000, the binder used for the surface dressing, has a Pen value of 80/100 when cut back with diesel fuel. Resurfacing with a single surface dressing using the same aggregate was carried out in late 1991. COST SAVINGS Potential savings in costs are considerable if marl is used instead of crushed stone for road base. The respective construction costs for the crushed limestone and marl road base in 1978 were about £17,500 and £6,000 per km. The cost of the stabilized section was not recorded but, assuming a cement price of £40 per tonne in 1978, 190 Woodbridge11 a cost of £15,000 per km has been estimated. The all-in construction cost of the Northern Highway was £78,000 per km. Cost savings were thus £1 1,500 per km for the unstabilized marl, or 15 percent of the construction cost. The difference was mainly attributed to the quarrying and additional haulage costs of the limestone. CONCLUSIONS 1. The selected marls do not conform to existing road-base spec- ifications in terms of their particle size distribution, Atterberg lim- its, and strength in soaked CBR tests. The particles are weak and the material breaks down further under compaction, resulting in the par- ticle size distribution becoming even finer. The fine material in the matrix is mostly silt sized, although the marl can contain clay lenses, which will increase the plasticity. 2. After trafficking for 19 years and cumulated axle loading of 1.3 million esa, deformation has been negligible. The good performance of the trial sections is attributed to the rigidity of the marl attained after compaction, as expressed by the compaction characteristics. How- ever, the marl is sensitive to increases in moisture content, which in the trial have been controlled by construction of an embankment and by maintaining the surface dressing in good condition. 3. The performance of the cement-stabilized marl section has been even better with development of very high strength and only minor shrinkage cracking. The success of the stabilization means that it may be possible to upgrade more plastic marls to a standard acceptable for road base. Better performance can be expected from the stabilized section than from the unstabilized marls because it will retain its strength under wetter conditions. 4. There has been virtually no deformation of the trial marl. sec- tions so it is difficult to predict their remaining life. From the analy- sis of structural number an additional life of between 1 million and 2 million esa is indicated. 5. The potential cost savings of using marls instead of crushed stone are significant. ACKNOWLEDGMENTS Throughout the project the Belize Ministry of Works provided gen- erous support and encouragement. BSI was kind enough to provide important data. Colleagues at TRL were involved in the earlier phases of the study and the engineers of the Crown Agents were instrumental in ensuring a high quality of construction. REFERENCES 1. Classification Tests~for British StandardMethodsof TestforSoilsfor Civil Engineering Purposes. BS 1377:. Part 2. British Standards Institution, 1990.2. Overseas Road Note 31: A Guide to the Structural Design of Bitumen- Surfaced Roads in Tropical and Sub-tropical Countries, 3rd ed. Transport and Road Research Laboratory, Crowthorne, Berkshire, U.K., 1977. 3. Overseas Road Note 31: A Guide to the Structural Design of Bitumen- Surfaced Roads in Tropical and Sub-tropical Countries, 4th ed. Transport and Road Research Laboratory, Crowthome, Berkshire, U.K, 1993. 4. Jenner, H. N., and R. H. Burfitt. Chalk: An Engineering Material. In Insti- tution of Civil Engineers Proceedings, 1974. 5. Liddle, W. J. Application of American Association of State Highway Offi- cials (AASHO) Road Test Results to the Design of Flexible Pavement Structures. In First International Conference on the Structural Design of Asphalt Pavements, University of Michigan, Ann Arbor, 1962. 6. Paterson, W. D. 0. Road Deterioration and Maintenance Effects. High- way Design and Maintenance Series, The John Hopkins University Press, Dec. 1987. 7. Hodges, J., J. Rolt, and T. E. Jones. The Kenya Road Transport Cost Study: Research on Road Deterioration. TRRL Laboratory Report 673, 1975.8. Rolt, J..An Improved Method of Calculating Pavement Structural Num- ber for Evaluation and Design. Working Paper No. 233. Overseas Unit, Transport and Road Research Laboratory, Crowthorne, Berkshire, U.K, 1987.9. Edwards, A. C., and R. J. Mead. Rotavapor Method for the Recovery of Bitumen. TRRL Working Paper No. 129, Pavement Materials and Construction Division, 1986. [. ,..t -, ..~~~~~~~~77 191