34 TRANSPORTATION RESEARCH RECORD 1291

Economic Viability of Upgrading Low-Volume

H. R. KERALI, M. S. SNAITH, AND R. C. KooLE

In many countries, particularly in the developing world, a large In many developing countries, a large percentage of the percentage of the network is made up of gravel roads. road network is unsealed (earth or gravel roads), and despite Although such roads are suitable for low traffic volumes, many the generally low traffic volumes on such roads, there is a countries have had rapid increases in traffic levels. This process need to protect them from being washed away, particularly has resulted in large expenditure on maintenance and rehabili­ tation of gravel roads as well as high costs of vehicle operation. in hilly or mountainous regions with high rainfall. The ingress At high traffic levels, it is therefore desirable to upgrade such of water can be prevented by sealing a using a roads with bitumen or pavements to reduce total costs. relatively cheap bituminous layer such as surface dressing or A computer package incorporating the World Bank other surface treatment. With increased traffic loading, a road Design and Maintenance Standards Model (HDM-III) has been requires additional strength to withstand the development of developed to determine the traffic levels at which it becomes permanent deformation in the subgrade and in other under­ economic to upgrade road pavements. Depending on the level of accuracy required, the computer package may be used to give lying granular layers. In order to reduce the stresses, an extra approximate or detailed results. At the simplest operation level, thickness of pavement is required. This is usually provided only a few data variables are required to accomplish a quick by gravel in sealed or surface-dressed roads, but even gravel analysis giving approximate results. A graphical plot of the total is becoming increasingly scarce and requires continual replen­ costs is displayed on the computer screen showing the crossover ishment. At a certain combination of gravel cost and traffic, point at which upgrading becomes viable. it becomes cheaper to add the required strength to the road in the form of a load-spreading structural bituminous layer. The purpose of constructing a road is to provide safe and A simplified method of economic appraisal has been devel­ reliable vehicle movement. In many countries, roads have to oped for calculating the costs and benefits of the transition be constructed along route corridors that pass through areas from unsealed roads to sealed roads and then further to roads with soils that are both weak and susceptible to the ingress with a structural bituminous load-bearing pavement layer. of water. The construction of an appropriate standard of pave­ The transition point is defined in terms of the number of ment is intended to provide a smooth riding surface in addition vehicles using a road per day required to justify a higher to protecting the underlying soil subgrade from the adverse pavement standard. This is referred to as the break-even traffic effects of traffic loading and excessive moisture. The function level for upgrading from a weak to a stronger type of pave­ of a highway engineer is therefore to design cost-effective ment. For a given set of circumstances, the transition point roads with appropriate geometric and pavement standards between pavement standards depends on the costs of road that meet the objectives. In order to perform this function, construction, road maintenance traffic composition, and the there are a number of different road types lying between resulting vehicle operating costs (VOCs). These costs will engineered gravel roads and heavy-duty bituminous or asphalt largely depend on the location of the project together with pavements that the highway engineer can adopt in the pave­ the effects of topography and climate. Several economic anal­ ment design stage, depending on the economic viability of the yses were conducted to determine typical break-even traffic proposed road. levels in a number of environments. The results are intended The progression from one of the above pavement standards for use in chart form to identify situations in which upgrading to the next represents a significant improvement in the capac­ a road pavement structure may be beneficial. ity of a road to withstand traffic loading and the adverse effects of the environment. An increase in construction expenditure SHELL INPUT DATA PROGRAM would be required, but this is countered by lower rates of pavement deterioration caused by traffic loading and the envi­ A user-friendly microcomputer program has been developed ronment. Each successive improvement in pavement stan­ to be used as a front-end data preparation tool. The primary dards may therefore be justified by consequent savings in objective of the Shell Input Data (SID) program is to prepare maintenance expenditure, in the cost of vehicle operation, data for the World Bank HDM III model in the form required and in other road user costs. Hence, it is clear that with high for the break-even analysis. HDM-III is then used to cal­ trnffic, or severe environment, there is a greater likelihood culate the transport costs incurred on any two pavement types that construction of a stronger pavement would be justified. at different traffic flow levels. The package therefore provides a user-friendly front end that can be used to run HDM-III H. R. Kerali and M. S. Snaith, School of Civil Engineering, Uni­ versity of Birmingham, Birmingham, B15 2TT, United Kingdom. with the specific purpose of determining the optimum traffic R. C. Koole, Shell International Petroleum Company, London, United levels for upgrading road pavements under different envi­ Kingdom. ronments. Kerali et al. 35

TABLE 1 TYPICAL DATA BANDS USED IN SID, LEVEL 2

Parameter Level 2 Default Values Low Average High

Environment: Rainfall (mm/month) 50 100 200 Altitude (mm) 200 1000 2000

Geometric Characteristics: Horizontal Curvature (deg/km) 50 180 500 Vertical Alignment (m/km) 5 10 30 Carriageway Width (m) 2.5 3.5 4.5

Pavement Standard: Surface Layer Thickness (mm) 25 50 100 Roadbase Thickness (mm) 100 250 500 Subbase Thickness (mm) 100 250 500 Subgrade CBR (%) 5 15 30

The SID program provides users with three modes of oper­ that, under the given set of circumstances, a sealed road has ation (Levels 2, 1, and 0), corresponding to the simplicity lower total transport costs and therefore its construction would required in data processing. In Level 2 operation, only a few be economically viable. parameters required to run HDM-III are requested; other In this example, the set of circumstances refers to a road input parameters are assigned default values. Data entry using with known climate, topography, and traffic flow character­ Level 2 is intended to give approximate results from which a istics. The method of appraisal could therefore be used to decision may be made for further investigation. In this mode, answer the question, Is it cheaper to operate a gravel or a only a few data cards can be edited to specify parameters such sealed road under the present traffic loading pattern? In order as the unit costs of construction, maintenance, vehicle oper­ to answer the next question-At what traffic flow level does ation, and traffic characteristics. Other physical characteristics it become cheaper to operate a sealed instead of a gravel of the road to be analyzed can be selected from low, average, road?-it is necessary to calculate the total transport costs or high values. Table 1 presents examples of default values for both road types for a range of traffic flow levels and to suggested by the SID program in Level 2 operation. The determine the average daily traffic (ADT) at which the trans­ default values are intended to give the orders of magnitude port costs for the two pavement types are equal. believed to be typical for the area of analysis and can be Weak pavements deteriorate faster than strong pavements changed by the user. However, Level 1 operation requests under traffic loading and other environmental effects. At low all parameters necessary to run HDM-III and is designed to traffic levels, the rate of pavement deterioration is low, and provide more accurate results. It is intended to be used by therefore transport costs are likely to be low. Consequently, experienced HDM-III users, who are familiar with all input by comparing the transport costs calculated for a range of requirements. Levels 1 and 2 are tailored to provide data to traffic loading on two pavement types, it is possible to inter­ HDM-III for the comparison of any two types of pavement polate and obtain the ADT at which transport costs are equal structure to obtain the break-even traffic level. Level 0 of the for the two pavement types. data input program can be used to specify input parameters The economic analysis required to determine the break­ to HDM-III for any type of economic analysis (e.g., the even traffic is conducted by using HDM-III to calculate trans­ appraisal of maintenance alternatives). port costs for any two pavement types under a range of traffic flow levels. Table 2 presents the default traffic flow levels used in HD M-III to compare transport costs calculated for ECONOMIC ANALYSIS OF BREAK-EVEN gravel, sealed, and premix roads. These values represent ini­ TRAFFIC LEVELS tial ADT with a specified traffic composition, that is, observed traffic flow in the first year of analysis (the base year). The The economic analysis involved in deriving the break-even traffic levels were chosen to include the upper and lower limits traffic level is a straightforward comparison of transport costs for the break-even ADT likely to be found in most countries. calculated for any two road types. For example, to determine When these are used in a particular country, the unit costs of whether a or a sealed road is more suitable under construction, maintenance, and vehicle operation are required a given set of circumstances, HDM-III is used to calculate together with the maintenance standards commonly applied. the annual transport cost matrices for the two alternatives. These unit costs, maintenance standards, and traffic mix affect For example, if the present value of costs for a gravel road the break-even ADT level. For countries in which costs of alternative exceeds that of a sealed road, it can be concluded vehicle operation are high, a low break-even ADT can be 36 TRANSPORTATION RESEARCH RECORD 1291

TABLE 2 TRAFFIC FLOW LEVELS USED IN HOM-III COMPARISONS

Alternative Gravel to Sealed Sealed to Premix Gravel to Premix

ALTl 20 100 100 ALT2 50 200 200 ALT3 75 400 400 ALT4 100 700 700 ALT5 200 1000 1000 ALT6 400 2000 2000 ALT7 700 5000 5000 ALTS 1000 10000 10000

expected. Similarly, if the maintenance costs for the weaker Presentation of Results pavement are high (e.g., cost of regraveling), a low break­ even traffic can also be expected. There are a number of ways in which the results given in Table 2 can be presented to illustrate the break-even traffic level. Three of these are of particular interest when presented Typical Results Using Data From Cyprus in graphical form: • Total transport costs versus initial ADT flows , Trial runs of the HDM-III program were conducted using • NPV versus initial ADT flows, and data from Cyprus with the objective of determining typical • Internal rate of return (IRR) versus initial ADT flows. break-even traffic flow levels for upgrading from gravel to sealed and from sealed to premix pavements. For these anal­ Figure 1 shows the first of these with the total transport yses, the gravel road was assumed to have 150 mm of gravel costs for the gravel and sealed roads on the vertical axis. This surfacing; the sealed road, 25 mm of surface dressing on 150 graph is similar to the graph produced on the computer screen. mm granular road base; and the premix road, 40 mm of premix The point at which the two curves intersect marks the break­ surfacing layer on top of 300 mm of selected granular road even ADT value. Figure 1 indicates that the break-even traffic base material. The results from the trial runs produced for a level is just under 200 vpd for a 2 percent annual traffic growth comparison of a sealed road against a gravel road by rate. Figure 1 can also be used to estimate the size of benefits HDM-III are summarized in Table 3. The break-even traffic to be derived by upgrading a road. If the daily traffic flow is level is given by the ADT level for which the net present value known, the benefits derived from upgrading are given by the (NPV) is zero (i.e., when the transport costs of the two pave­ difference in total transport costs. ments being compared arc equal). It may be observed from Figure 2 shows the sewml method of presentation with the the table that the optimum or break-even traffic level required NPVs at four discount rates plotted on the vertical axis. The to justify the construction of a surface-dressed road varies break-even traffic level for each of the discount rates is given with the discount rate used in the analyses. If the economic by the points at which the curves cross the horizontal x-axis costs are not discounted, the break-even traffic flow is just when the NPV i~ Lt;JlJ. Tlte 10 µe1ct::11l uiswu11l line C!Usses over 50 vehicles per day (vpd). the x-axis at a traffic flow of just under 100 vpd.

TABLE 3 COMPARISON OF A SEALED ROAD WITH A GRAVEL ROAD FOR 2 PERCENT TRAFFIC GROWTH RATE

Initial Net Present Value at Discount Rate of I.RR Traffic 0% 5% 10% 15% (%)

20 -0.215 -0.217 -0.218 -0.219 -20.7 50 -0.134 -0.163 -0.179 -0.189 -8.6 75 -0.064 -0.119 -0.149 -0.166 -3.3 100 0.024 -0.066 -0.114 -0.141 1.1 200 0.296 0.107 0.007 -0.051 10.5 400 1.362 0.754 0.435 0.254 32.6 700 3.527 2.081 1.322 0.891 71.4 1000 5.817 3.502 2.281 1.584 109.2 Kerali et al. 37

Cost (C£m) 7 Traffic Growth - 2% 6 Discount Rate - 10%

5 Sealed ... ------,,.,-­ 4 _,., ,-' 3 ------~---~------2

0 0 100 200 300 400 500 600 700 800 900 1000

ln~ial Average Daily Traffic (ADl) FIGURE 1 Relationship between total transport cost and traffic level.

Table 3 presents IRR values calculated for each of the initial tenance in most cases are determined before the analysis is ADT flows . The IRR values represent the discount rates at conducted and are then assumed to remain constant. The which the corresponding ADT would justify upgrading from break-even traffic level therefore depends largely on traffic a gravel to a sealed road. Figure 3 shows the relationship growth and the discount rates used in the economic analyses. between IRR and traffic flow for Cyprus conditions. The If the voe component is included in the analyses, the size figure represents all ADT levels required to justify upgrading of the total saving in voe will depend on the traffic growth a gravel road to a surface-dressed road in Cyprus for any rate. If high traffic growth rates are used, the traffic loading given traffic growth and discount rate. For example, at a during the analysis period will be high, resulting in a faster discount rate of 10 percent, the break-even traffic flow for deterioration of the weaker pavement and giving higher sav­ surface dressing is just under 200 vpd for a 2 percent growth ings in VOC; consequently, upgrading to a stronger pavement in traffic. will be justified at lower traffic flow levels. Figure 3 shows the effects of these factors on the break-even traffic levels. Factors Affecting the Break-Even Traffic Level

The break-even traffic level discussed in the previous section EFFECT OF ENVIRONMENT ON BREAK-EVEN depends on a number of factors, notably the unit costs of TRAFFIC LEVELS construction and maintenance, traffic growth rates, discount rates, and the traffic composition that has a direct effect on Economic analyses were conducted using HDM-III to deter­ the VOC component. The unit costs of construction and main- mine the break-even traffic for a range of environments using

NPV (C£m) 20 Discount Rate 18 0% 16 14 12 10 _,-· 5% .,.,-;--- ...... - 8 __ ... ------6 10% 4 15% 2 - -- ...... 0 -2 100 200 300 400 500 600 700 800 900 1000 Initial Average Daily Traffic (ADT) FIGURE 2 Relationship between NPV and traffic level. 38 TRANSPORTATION RESEARCH RECORD 1291

IRA% Traffic Growth 50 10%

40 ... - 5%

----·- - ·- ....- · 2% 30 .... _.. __ ..... --- - ... -.-· ...... - _____ . ..- ·· 20 -·- ..-·- 10

150 200 250 300 350 400 -10 Initial Average Daily Traffic (ADT)

-20

-30 FIGURE 3 Relationship between IRR and traffic level.

data from Cyprus. The objective for this procedure was to to predicted cumulative traffic loading and prevailing subgrade quantify the effects of various environmental factors on the strength conditions over the design period. The break-even break-even traffic levels. The factors included in the analysis analysis is essentially conducted by calculating the total trans­ were rainfall, topography, and subgrade soil strength in terms port costs incurred on a road using a range of traffic flow of the California bearing ratio (CBR). In the course of the levels. Ideally, each traffic range used in the analysis should analysis, it was found that some of the environmental factors be assigned a unique pavement design (and hence cost of had little or no effect on the break-even traffic results (e.g., construction). However, to maintain the simplicity of the cal­ rainfall on sealed or premix roads). culation of the break-even traffic level, the same pavement The Shell pavement design manual recommends pavement structure is used in all analyses. This pavement structure layer thicknesses based on four factors: represents the minimum possible design thickness. For exam­ ple, the minimum pavement structure for premix roads is • Traffic loading in terms of equivalent standard axles, assumed to include a 50-mm asphalt (or equivalent) • Subg1aJe sl1eugll1 muJulu~, surfacing and 300 nun of granular road_ base. The minimum • Prevailing weather conditions in terms of the weighted sealed road structure is a surface dressing of 25 mm with a mean temperature, and granular base of 300 mm. For unpaved roads, the minimum • Bitumen and mix properties. gravel thickness commonly used is 150 mm. This assumption is compensated for in the analyses by using The effect of traffic loading and subgrade strength on the a condition-responsive maintenance policy within HDM-111. design thickness for pavements used in the break-even anal­ The effect is to apply frequent maintenance and rehabilitation ysis is discussed in the next section. The effect of temperature on undcrdcsigncd pavements when the pavement attains crit­ on the performance of bituminous pavements is not explicitly ical condition. For example, a pavement composed of 50 mm modeled within HDM-111. However, temperature variations of and 300 mm of granular road base is used in different climatic regions can be taken into account by for ADT levels from 100 to 5,000 but is expected to receive calibrating the performance equations built into the model, more overlays during the analysis period under the heavier such as depth progression and other defect progressions. traffic load. The effects of the environment are also taken into account in terms of the rainfall and altitude. In HDM-111, rainfall is Effect of Rainfall and Topography assumed to affect the performance of unpaved roads, partic­ ularly the rate of gravel loss. Only rainfall in combination with high vertical alignment had noticeable effects on the break-even traffic level. The analyses were therefore performed for a combination of low rainfall Pavement Design Standards in a flat area and high rainfall in a mountainous area. The results obtained for intermediate combinations of rainfall with Traffic loading is used in the break-even analysis as the inde­ topography indicated deviations in the break-even traffic level pendent variable that governs the size of benefits derived from of less than 20 vpd. upgrading a road from one pavement type to another. The The break-even traffic levels for upgrading roads from gravel normal practice is to design pavement thicknesses according to sealed pavements range from 50 to 150 vpd depending on 60 Annual Traffic MODERATE RAINFALL Growth LOW LYING AREA

50

,.-., 5% ._,i!R 2% i:l 40 ....s llJ ....~ 0 30 ....Ill QI ~ Cl i:l 20 ....Ill"' ....i:l

10

0 -1-~~~.--~~~.--~~--r~~~-r~~~-.-~~~...-~~--.~~~~~~~~ 0.1 0.3 0.5 0.7 0.9 (Thousands) Average Daily Traffic (.ADT)

Annual Traffic Growth 45 -.-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.... 1oi HIGH RAINFALL MOUNTAINOUS AREA 40 5%

35 2% ,-.. ._,i!R i:l 30 ....::l"' Ill ~.... 25 0 ....II) QI 20 ~ QI i:l 15 ~"' ...i:l 10

5

0 50 150 250 350 450 Average Daily Traffic (.ADT)

FIGURE 4 Effect of rainfall and topography on break-even traffic level: (top) upgrading gravel to premix road and (bottom) upgrading gravel to sealed road. 40 TRANSPORTATION RESEARCH RECORD 1291 the traffic growth rate and the discount rate used in the anal­ at which the transition from one pavement standard to the ysis. The combined effect of rainfall and severe topography next becomes economic. A number of analyses were con­ is to lower the break-even traffic level by between 20 to 50 ducted using the package to study the effects of the climate, vpd, again depending on the traffic growth rate and the dis­ topography, and sub grade soil strength on the break-even count rate. Figure 4 shows the results of break-even analyses traffic levels. The results of the analyses indicate that the conducted for a low-lying area with moderate rainfall and for break-even traffic level for upgrading roads from gravel to a mountainous area with high rainfall. sealed is surprisingly low, implying a rapid increase in trans­ The comparison of sealed-road performance against that of port costs once the break-even traffic level for paving a road premix roads gave some unexpected results, because the break­ is exceeded. This confirms the view that sealing a road at an even traffic level from sealed to premix was between 500 to early stage is advisable. The combined effects of heavy rainfall 750 vpd. This comparison refers to premix roads with the and severe topography result in a lower break-even traffic minimum pavement structure (i.e., 50 mm asphalt concrete level than might be expected. The comparisons of sealed-road on 300 mm granular road base). This result is in contrast with performance and that of premix roads also gave some unex­ ADTs of over 1,500 vpd, commonly quoted as the minimum pected results, because the break-even traffic level from sealed for premix standard pavements. to premix occurs at much lower ADT flow levels than is The break-even traffic level for upgrading from a gravel to commonly expected. a premix rn;ici is surprisingly low, ranging from 100 to 250 vpd. This result indicates a rapid increase in transport costs once the break-even traffic level for paving a road is exceeded. ACKNOWLEDGMENTS It therefore suggests that sealing a road at an early stage will yield high benefits. The work described in this paper was sponsored by Shell The preceding results point to a high degree of dependence International Petroleum Company (SIPC). The Highways and on the unit costs of construction used for gravel roads, sealed Geotechnics Research Group of the University of Birming­ roads, and premix roads together with the unit costs for main­ ham is greatly indebted to SIPC for providing the resources tenance activities. Gravel roads, though least expensive to required in the research study. construct, have a high voe component because of high sur­ face roughness values. The difference in roughness progres­ sion on sealed and premix roads, however, is similarly high, REFERENCES resulting in a lower break-even traffic level than is commonly expected. 1. World Bank. The Highway Design and Maintenance Standards Model. Johns Hopkins University Press, Baltimore, Md., 1988. 2. The SHELL Pavement Design Manual for Asphalt Pavements and CONCLUSIONS Overlays for Road Traffic. Shell International Petroleum Com­ pany, London, 1978. 3. A Guide to Road Project Appraisal. Overseas Road Note 5. Trans­ A computer package incorporating the World Bank's port and Road Research Laboratory, Crowthorne, Berkshire, HDM-III model can be used to determine the traffic levels England, 1988.