Civl061203.qxd The Journal of Engineering Research Vol.2, No. 1 (2005) 32-42 ___________________________________________ *Corresponding authors E-mail: hossam@squ.edu.om Development of Asphalt Pavement Temperature Models for Oman H.F. Hassan*1, A.S. Al-Nuaimi1, R. Taha1, and T.M.A. Jafar2 1Department of Civil and Architectural Engineering, Sultan Qaboos University, P.O. Box 33, Al Khod 123, Muscat Sultanate of Oman 2 Research and Technical Studies Department, Muscat Municipality, Muscat, Sultanate of Oman Received 6 December 2003; accepted 6 June 2004 Abstract : Asphalt pavements form an integral part of any transportation system. The structural capacity of the hot mix asphalt concrete layers depends on many factors including its temperature. Moreover, temperature can be a major con- tributor to several types of distresses. Therefore, temperature is a significant factor that affects the performance and life span of a pavement. The Sultanate of Oman's road network expanded at a phenomenal pace from approximately 10 km of paved roads in 1970 to 9,673 km in 2001. with the recent SHRP and LTTP research findings, it was necessary to inves- tigate the applicability of the models developed from these research studies to Oman's environmental conditions and more generally to the Arabian Gulf climate. This paper presents the research undertaken to develop models to predict high and low asphalt pavement temperatures in Oman. A pavement monitoring station was set-up at the Sultan Qaboos University (SQU) campus to monitor air, pavement temperatures and solar radiation. Data were collected for 445 days. Daily min- imum and maximum temperatures were recorded. A regression analysis was used to develop the low pavement temper- ature model. A stepwise regression was used to develop high temperature models using air temperature, solar radiation, and duration of solar radiation as independent variables. The developed models were compared with the SHRP and LTPP models. The SHRP and LTPP models were found to be more conservative than the developed models, which are more suitable for predicting pavement temperatures in Oman, and more generally in the Gulf region. Keywords: Pavement temperature, Models, SHRP, LTPP, Asphalt concrete, Oman ¢¢üü∏∏îîàà°°ùŸŸGGGòg .IQGô◊G áLQO ɡ檰V øe πeGƒY I~Y ≈∏Y á«à∏Ø°SC’G áfÉ°SôÿG ∞°UQ á≤ÑW πª– Iƒb ~ªà©Jh ,π≤f áµÑ°T iCG ‘ »°ù«FQ AõL á«à∏Ø°SC’G ¥ô£dG Èà©J : ¥ô£dG áµÑ°T äQƒ£J ~bh Gòg . ∞°UôdG äÉ≤ÑW ôªYh AGOCG ≈∏Y kGôKDƒe kÓeÉY IQGô◊G Èà©J ∂dòdh .∞°UôdG äÉ≤ÑW ܃«Y øe áØ∏àfl ´GƒfCG Qƒ¡X ≈∏Y IQGô◊G ÒKÉàd áaÉ°VE’ÉH Ωɶf Qƒ£Jh Qƒ¡X ~©Hh .2001 ΩÉY ‘ ºc 9700 ‹GƒM ¤EG 1970 ΩÉY ‘ áaƒ°UôŸG ¥ô£dG øe ºc 10 ‹GƒM øe IÒÑc ä’~©Ã ¿ÉªY ‘ SHRPèeÉfôH h LTPP¿ÉªY ±hô¶d êPɪædG √òg áÑ°SÉæe …~e ~j~–h ΩɶædG Gòg ∫ÓN øe äQƒW »àdG êPɪædG ≥«Ñ£J øe ≥≤ëàdG øe ~H’ ¿Éc âjôLG »àdG çÉëHC’G èFÉàf Qƒ¡X h ∞°UôdG äÉ≤Ñ£d …ô¨°üdG h iƒ°ü≤dG √QGô◊G äÉLQO ™bƒàd êPɉ OÉéjE’ √DhGôLEG ” …òdG åëÑdG èFÉàf ábQƒdG √òg ¢Vô©à°ùJh .»Hô©dG è«∏ÿG á≤£æe ñÉæŸ ºYCG áØ°üHh á«NÉæŸG ~j~–h Ωƒj 445 I~Ÿ äÉfÉ«ÑdG ~°UQ ” ~bh .»°ùª°ûdG ´É©°TE’G ∂dòch ∞°UôdG äÉ≤ÑWh ƒ÷G IQGôM äÉLQO ¢SÉ«≤d á£fi Ö«côJh ™bƒe õ«¡Œ ” ~bh .¿ÉªY ‘ »à∏Ø°SC’G êPƒ‰ OÉéjE’ äGƒ£N hP …QG~ëfE’G π«∏ëàdG á≤jôWh ,iô¨°üdG √QGô◊G äÉLQ~d êPƒ‰ OÉéjE’ …QG~ëfE’G π«∏ëàdG á≤jôW ΩG~îà°SEG ”h ,á«eƒ«dG …ô¨°üdGh …ƒ°ü≤dG IQGô◊G äÉLQO á°SGQ~dG √òg ‘ IQƒ£ŸG êPɪædG áfQÉ≤e ” ~bh .á∏≤à°ùe äGÒ¨àªc »°ùª°ûdG ´É©°TE’G √Îa ∂dòch »°ùª°ûdG ´É©°TE’Gh ƒ÷G IQGôM áLQO ΩG~îà°SEÉH iƒ°ü≤dG IQGô◊G äÉLQ~d êPɉ ™e SHRPhLTPPIQGôM äÉLQO ™bƒàd áªFÓe ÌcCG Èà©J »àdGh á°SGQ~dG √òg ‘ IQƒ£ŸG êPɪædG øe kɶؖ ÌcCG IÒNC’G êPɪædG ¿G áfQÉ≤ŸG øe ÚÑJ h .è«∏ÿG á≤£æe ‘ áeÉY áØ°üHh ¿ÉªY ‘ »à∏Ø°SC’G ∞°UôdG áá««MMÉÉààØØŸŸGG ääGGOOôôØØŸŸGG.¿ÉªY ,á«à∏Ø°S’G áfÉ°SôÿG ,ÜQÉ°T ,êPɉ ,∞°UôdG IQGôM áLQO : ¿ÉªY ‘ »à∏Ø°S’G ∞°UôdG IQGôM äÉLQ~d êPɉ ôjƒ£J 33 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 1. Introduction Asphalt pavements form an integral part of any trans- portation system. A significant factor that affects the per- formance and life span of a pavement is the influence of temperature. Temperature can contribute to certain com- mon types of asphalt pavement distresses such as perma- nent deformation or rutting (typically associated with high temperature environments), bleeding, and thermal crack- ing (associated with low temperature environments). The Strategic Highway Research Program (SHRP) established the Long Term Pavement Monitoring Program (LTPP) program in 1987 to support a broad range of pave- ment performance analyses leading to improved engineer- ing tools to design, construct, and manage pavements Diefenderfer, et al. (2002). The Seasonal Monitoring Program (SMP) was estab- lished as an element of LTPP in 1991 to measure and eval- uate the effects of temperature and moisture variations on pavement performance and validate the available models (Mohesni, 1998; Diefenderfer, et al. 2002). From the initial SHRP testing and SMP data, several pavement temperature models were developed to assist in the proper selection of the asphalt binder performance grade (Mohesni, 1998; Mohesni and Symons, 1998a; Mohesni and Symons, 1998b; Lukanen, 1998; Diefenderfer, et al. 2002). Solaimanian and Kennedy (1993) proposed an analytical model based on the theory of heat and energy transfer. Shao et al. (1997) also devel- oped a procedure based on heat transfer theory to estimate pavement temperatures. Regression models based on other sets of data were developed (Bosscher et al. 1998; Marshall et al. 2001). A simulation model was developed to calculate temperatures during summer conditions based on the heat transfer models developed by Solaimanian and Kennedy (Hermansson, 2000 and 2001; Diefenderfer, et al. 2002). Al-Abdul Wahhab et al. (1994) conducted a study in two regions in Saudi Arabia to manually measure pave- ment temperatures in different pavement sections. The study concluded that the extreme pavement temperatures in arid environment ranged between 3 and 72°C, while in coastal areas, the temperature ranged between 4 and 65°C. In another study, Al-Abdul Wahhab et al. (1997) recom- mended five performance graded binder zones for the whole Gulf area. The study also proposed modification of the currently used binders to suite the proposed grades. The Sultanate of Oman as well as the Gulf area, in gen- eral, possesses a different environment from that of the North America. The applicability of the recent SHRP and LTPP developments in the US to Oman's or the Arabian Gulf's environmental conditions needed to be evaluated. This paper presents the research undertaken to develop models to predict high and low asphalt pavement temper- atures in Oman. A pavement monitoring station was estab- lished at the Sultan Qaboos University (SQU) campus to monitor air, pavement temperatures, and solar radiation. Data was collected for 445 days. Daily minimum and maximum temperatures were recorded. A regression analysis was used to develop the low pavement tempera- ture model. A stepwise regression was used to develop high temperature models using air temperature, solar radi- ation, and duration of solar radiation as independent vari- ables. The instrumentation used is described and collect- ed data are presented. The developed models were com- pared with the SHRP and LTPP models. Notation df = degree of freedom, F = F statistic, Lat. = latitude, MS = mean sum of squares, P-value = power of the test, R2- adj. = adjusted coefficient of determination R2 = coefficient of determination, SEE = standard error of estimate, Solar = cumulative solar radiation from dawn to the time of occurrence of maximum air tempera- ture , W.hr/m2, SS = sum of squares, Std. Error = standard error or standard deviation of the coefficient's distribution, Tair = air temperature, °C, trad = duration of sunshine from dawn to the time of occurrence of maximum air temperature, hours, t-Stat = t statistic, Tsurf = pavement surface temperature, °C, α = level of significance. 2. Pavement Temperature Monitoring Station A monitoring station was set-up to collect data on air temperature, solar radiation and pavement temperatures at various depths. Sensors were connected to a CR10x Campbell Scientific datalogger housed in a protection enclosure. The enclosure was mounted on an instrumen- tation tower. The datalogger was operated by a battery. The following section describes the installation and sensor locations. 2.1. Installation of Station A test area 7x8 m was selected for the station installa- tion inside a newly built parking lot at the Sultan Qaboos University campus. The pavement cross section was mod- ified from the original design in the area of the test section to be composed of 135 mm of asphalt concrete (AC) wear- ing course on top of 165 mm of unbound base course on top of compacted subgrade soil. After the construction of the pavement was completed, the installation of the sensors was initiated. The procedure started with the excavation of a 500 mm by 500 mm by 480 mm deep pit. A trench 200 mm wide by 480 mm deep was also excavated from the pit to where the tower was installed. A fence was erected around the tower area to protect the equipment from damage. A 50 mm PVC pipe was laid in the trench to protect the sensor cables extending from the pit to the tower location. All sensors were inserted into the PVC pipe and led to the excavation pit providing adequate slack so as to allow for comfortable installation of the probes at the desired loca- tions. Two sets of thermistors referred to as center and right, were installed in the center and right sides of the pit. The right set consisted of eight sensors covering the full depth of the excavated pit. The center set had four sensors that covered only the depth of the AC layer to serve as a dupli- cate for the first set. At the targeted sensor locations, a mark was made at the appropriate side of the pit. An elec- trically powered horizontal and 90° angle hammer drill was used to drill holes approximately 280 mm deep into the side of the excavated pit. The 90° angle was used to accommodate the hammer drill and drilling bit inside the limited pit space. The process was repeated for each sensor with the exception of those located at the surface (at 0 mm depth). Holes were cleared of loose debris before sensors were inserted. Sensors were then inserted all the way into the drilled holes and sealed in place using a silicon sealant to prevent concrete or moisture infiltration when backfilling the pit. Figure 1 shows the sensors inserted in the drilled holes at the desired depths. For surface temperature sensors, a grinder was used to cut through the pavement surface such that the probe could be flush with the pavement surface. The probe was positioned in place and a cold sand asphalt mix was used to seal around it. All actual depths of pavement sensors were measured and recorded. Figure 2 shows a schemat- ic representation of the installation pit showing the layer thicknesses and the actual depth of the sensors. After adjusting the extra cable lengths, the pit and trench were filled with the subgrade and base material. Compaction of the material was performed manually. Portland cement concrete was used to backfill the pit up to the surface level instead of asphalt concrete due to the ease of backfilling with concrete in addition to the fear of dam- aging the sensor cables from the hot mix. It was felt that using the concrete instead of asphalt concrete will not affect the readings from the probes, since the sensing tip of the probe (60 mm long) was buried inside the asphalt con- crete side of the pit at least 220 mm away from the inter- face between the existing pavement and poured concrete. Finally, the tower was erected and the datalogger solar radiation sensor and air temperature thermistor were mounted on the tower. The sensors were connected to the datalogger. Figure 3 shows an overview of the test area 34 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 Figure 1. Installation of thermistors 35 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 Section of Instrumentation Hole (Right) (Center) Section - Location of Sensors Section - Location of Sensors Plan of Instrumentation Hole (Dimensions are in mm) Figure 2. Pavement layers and thermistor locations Figure 3. General view of experimental test area Pit and Trench Tower and data logger cabinet after the set-up of the station. A program was written to operate the CR10x datalog- ger based on the available sensors configuration. The reading from the solar radiation sensor is taken every minute, and the average value is stored every 15 minutes. All other sensors are read every 15 minutes. Data are stored on the datalogger and downloaded to a computer via a storage module. 3. Data Analysis The temperature data collected were presented to show the different variability patterns between air and pavement temperatures. Pavement temperature models were estab- lished using regression analysis on the collected daily maximum and minimum temperatures. The data used to 36 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 15 20 25 30 35 40 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 Time of Day T em pe ra tu re , o C Air Temperature Surface 20 mm 65 mm 125 mm 210 mm 290 mm 375 mm Figure 4. Air and pavement temperature variation on the 24th and 25th December, 2002 20 25 30 35 40 45 50 55 60 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 Time of Day T em pe ra tu re , o C Air Temperature Surface 20 mm 65 mm 125 mm 210 mm 290 mm 375 mm Figure 5. Air and pavement temperature variation on the 2nd and 3rd of May 2003 develop the models were collected in the period from 23rd of December 2002 to 6th of April 2004, with a total of 445 days. 3.1. Temperature Data Figures 4 and 5 show the variation of air and pavement temperatures at different depths with the time of day. Fig. 4 shows data obtained for two consecutive days during the cold season; namely the 24th and 25th of December, 2002, while Fig. 5 shows this variation for two consecutive days during the hot season; namely the 2nd and 3rd of May, 2003. It can be seen that thermistors placed at 290 mm depth or more (located at the boundary of the subbase and sub- grade layers) are less sensitive to changes in the air tem- perature than sensors located from the surface down to 210 mm deep. The surface layer is the layer most sensi- tive to temperature fluctuations. There is a noticeable time lag in the heating cycle between the maximum air temper- ature and pavement temperature. In other words, it takes more time to heat the pavement than the air. The time lag increases as pavement depth increases. Time lag is also noticed between the minimum air temperature and pave- ment temperatures in the cooling cycle. The pavement surface cools quicker than other deeper layers. Figures 4 and 5 also indicate that the maximum or min- imum pavement temperatures occur some time after the occurrence of maximum or minimum air temperature, respectively. The minimum pavement temperature is recorded at the surface but is generally warmer than mini- mum air temperature. This agrees with the findings reported in the literature. Figures 6 and 7 show the same data only for the air, 37 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 10 15 20 25 30 35 40 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 Time of Day Te m pe ra tu re , o C 0 50 100 150 200 250 So la r R ad ia tio n, W .h r/m 2 Air Temperature Surface 20mm depth Solar Radiation 20 25 30 35 40 45 50 55 60 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 20 0 40 0 60 0 80 0 10 00 12 00 14 00 16 00 18 00 20 00 22 00 0 Time of Day T em pe ra tu re , o C 0 50 100 150 200 250 300 So la r R ad ia tio n, W .h r/m 2 Air Temperature Surface 20mm depth Solar Radiation Figure 6. Temperature and solar radiation for 24th and 25th of December, 2002 Figure 7. Temperature and solar radiation for 2nd and 3rd of May, 2003 surface and 20 mm sensors, in addition to the solar radia- tion. Fig. 4 illustrates some fluctuation in temperatures detected by the air temperature sensor and the sensors placed at the surface and 20 mm depth. This is more clear- ly illustrated in Fig. 6 where the readings obtained for solar radiation as well as the readings from the surface and 20 mm sensors are plotted. This effect is not seen in hot- ter conditions (Fig. 7). This can be explained by the cloudy weather conditions encountered in winter that shade the sun's rays, as shown by the fluctuation in solar radiation readings presented in Fig. 6. It can also be seen that the temperature of the pavement at greater depths is warmer than those placed nearer to the surface at night time. This pattern is maintained until the minimum air temperature is reached, and the pattern is reversed with the surface layers becoming warmer and gain energy from the sun rays during the day while the deeper layers become cooler. 3.2. Development of Low Temperature Model The low pavement temperature model was established using air temperature as an independent variable and pave- ment surface temperature as the dependant variable. The lowest air temperature in any single day was identified for the entire period during which data were collected and the corresponding daily minimum pavement temperature at the surface was obtained. A regression analysis was per- formed using Minitab statistical software to develop the model (Neter et al., 1990; MINITAB, 2000). The model was developed using 445 data points. Tsurf = -1.437 + 1.121 Tair (1) where, Tsurf = minimum pavement temperature, ºC and Tair = minimum air temperature, °C. This relationship with the collected data is illustrated in Figure 8. The model achieved a Goodness-of-Fit (R2) of 0.968 and a standard error of estimate (SEE) of 1.11 °C. Table 1 shows the ANOVA results for the model. It is to be noted that the lowest pavement temperatures recorded in Oman are well above the low pavement temperatures typically encountered in the US. This is due to the Gulf area climate which does not have subfreezing tempera- tures. 3.3. Development of High Pavement Temperature Models The highest air temperaure and the highest pavement temperature at a depth of 20 mm below the surface for each day were identified. A linear regression was per- formed using the highest air temperature as the independ- ent variable and the highest 20 mm pavement temperature as the dependent variable. Using 445 days of collected data, the following equation was obtained: 38 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 Tsurf = -1.437 + 1.121 T air R2 = 0.968 10.0 15.0 20.0 25.0 30.0 35.0 40.0 10.0 15.0 20.0 25.0 30.0 35.0 40.0 Air temperature, oC Pa ve m en t s ur fa ce te m p. , o C Figure 8. Developed low pavement temperature prediction model df SS MS F Regression 1 16155 16155 13150.93 Residual 443 544 1 Total 444 16699 Coefficients Std. Error t-Stat P-Value Intercept -1.4373 0.239 -6.02 0.000 Variable 1.1208 0.010 114.68 0.00 0 Table 1. ANOVA table for low temperature model* *df = degree of freedom; SS = sum of squares; MS = mean sum of squares, F = F statistic, Std.Error = standard error or standard deviation of the coefficient’s distribution T20mm = 3.160 + 1.319 Tair (2) where, T20mm= pavement temperature at 20 mm depth, ºC and Tair = maximum air temperature, ºC. Figure 9 shows the data collected as well as the regres- sion equation obtained. The equation achieved an R2 of 0.847 and an SEE of 3.18°C. Table 2 shows the ANOVA results for the model. Since high pavement temperature is affected by factors other than air temperature, other parameters were consid- ered with the aim of improving R2 of the model. Three variables were considered as independent variables: air temperature (Tair), cumulative solar radiation from dawn to the time at which the maximum air temperature occurred (Solar), and the duration over which solar radia- tion is accumulated (trad). The pavement temperature at 20 mm depth was taken as the dependant variable. In order to obtain the best model, a stepwise regression was per- formed. The accumulated solar radiation was considered as this is the energy that the pavement absorbs and con- tributes to the highest pavement temperature. The dura- tion to the maximum pavement temperature would have been more appropriate to consider, however, the identifi- cation of the maximum pavement temperature would not be possible at other weather stations where no pavement sensors would be available. The addition and removal of variables were allowed in the stepwise regression procedure. The Partial F-test was used to verify for the entry and removal of variables from the model. The stepwise regression resulted in one addi- tional model. Thus, a second model was developed using pavement temperature as the dependent variable, and air temperature and cumulative solar radiation as the inde- pendent variables. The developed equation was as fol- lows: T20mm = 2.713 + 1.281 Tair + 0.00053 Solar (3) where, Solar = cumulative solar radiation from dawn to the time of occurrence of maximum air temperature in W.h/m2. R2 increased and SEE decreased slightly com- pared to the first model (Eq. 2) to 0.854 and 3.12°C, respectively. 39 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 T20mm = 3.160 + 1.319 T air R2 = 0.847 20.0 30.0 40.0 50.0 60.0 70.0 80.0 15.0 20.0 25.0 30.0 35.0 40.0 45.0 50.0 Air temperature, oC Pa ve m en t t em p. , o C Figure 9. Developed high pavement temperature prediction model df SS MS F Regression 1 24512 24512 2416.44 Residual 435 4413 10 Total 436 28925 Coefficients Std. Error t-Stat P-Value Intercept 3.160 0.907 3.48 0.001 Variable 1.319 0.0268 49.16 0.000 Table 2. ANOVA table for high temperature model with air temperature variable* * Same notations as in Table 1 Table 3 shows the results of the stepwise regression. The analysis used the same level of significance (0.05) to add or remove variables. Using the partial F-test, if the P value is less than 0.05, then there is enough evidence that the variable is not zero at 0.05 level of significance and the variable should be kept. Otherwise, the variable should be removed. The table shows that the P value is less than 0.05 in both models. The third parameter (trad) was removed from the model as the P value was more than 0.05. A correlation between the cumulative solar radiation and the duration (trad) vari- ables was found to be 0.839 as expressed by the Pearson product moment correlation coefficient. This shows a strong correlation between the two variables, and supports the removal of the (trad) variable. The use of the developed relationships depends on the availability of data. Equation 2 can be used if only air temperature measurements are available. The more accu- rate model Eq. 3 can be used if solar radiation measure- ments are available. It should be noted that solar radiation may be estimated using the relationship described in details by Diefenderfer et al. (2002) which estimates solar radiation using the geographical latitude of any given location and day of the year. 4. Comparison of Developed Models with SHRP and LTPP Models 4.1. Low Temperature Model Figure 10 shows a comparison between the developed low temperature model and SHRP and LTPP low temper- ature prediction models for pavement surface temperature. The collected field data are also shown in Fig. 10. The SHRP and LTTP low temperature models are given by the following relationships, respectively Mohseni, (1998): Tsurf = 0.859Tair + 1.7 (4) Tsurf = 1.56 + 0.72Tair - 0.004Lat2 + 6.26 log10 (25) (5) 40 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 Step 1 2 Constant 3.160 2.713 Tair 1.319 1.281 t-stat. 49.16 46.14 P-Value 0.000 0.000 Solar 0.00053 t-stat. 4.32 P-Value 0.000 SEE 3.18 3.12 R2 84.74 85.37 R2 adj. 84.71 85.31 Table 3. Stepwise regression analysis for high pavement temperature model σ for entry = 0.05, α to remove = 0.05 10 15 20 25 30 35 40 10 15 20 25 30 35 40 Air temperature, oC Pa ve m en t s ur fa ce te m p. , o C LTPP SHRP Low temp. model Figure 10. Comparison between low pavement temperature prediction model and SHRP and LTTP models where, Tsurf = pavement surface temperature, Tair = air temperature, and Lat=latitude. The latitude was substitut- ed with the Seeb International Airport (Muscat) latitude, which is 23° 35' 00". Figure 10 clearly shows that the predicted SHRP and LTPP low pavement temperature is lower than both the measured values and the low pavement temperature pre- dicted by the developed model. Both models tend to be closer to both the collected data and the developed model at temperatures less than 15°C. This indicates that the developed model is more suitable for Oman's climate and that SHRP and LTPP models would be more conservative in the Performance Grade (PG) binder selection. 4.2. High Pavement Temperature Models Figure 11 shows a graphical comparison between the developed high pavement temperature prediction model, SHRP and LTPP models. The collected data are also pre- sented. SHRP and LTPP models are given by the follow- ing relationships, respectively Mohesni, (1998): T20mm = (Tair - 0.00618Lat2 + 0.2289Lat + 42.4) 0.9545 - 17.78 (6) T20mm = 54.32 + 0.78Tair - 0.0025Lat2 - 15.14log10(45) (7) where, T20mm = pavement temperature at 20 mm depth, Tair = air temperature, and Lat=latitude. The latitude was replaced with Muscat latitude. Figure 11 illustrates that the predicted SHRP and LTPP high pavement temperature at 20 mm are higher than both the measured values and the high pavement temperature predicted by the developed model. Therefore, the devel- oped model is more representative of Oman's climatic conditions. SHRP and LTPP models would be expected to result in a more conservative selection of the PG binder. 5. Conclusions The following conclusions can be drawn from the analysis of the data collected in this research: 1. A pavement temperature monitoring station was set-up on Sultan Qaboos University campus. The collected data indicated the expected trends for cycles of heating and cooling of the pavement layers. 2. A low pavement temperature model was developed for Oman based on air temperature as an independent vari- able. The model achieved an R2 of 0.968. 3. The low temperature model, when compared with SHRP and LTPP, showed that the two later models pro- vide lower estimate of pavement temperature compared with the developed model. SHRP and LTPP models are more conservative in the selection of Performance Grade (PG) binders, and the developed model would be more suitable for Oman's climate. 4. Two high pavement temperature models were devel- oped, taking into account air temperature and cumula- tive solar radiation from dawn to the time of maximum air temperature as independent variables to predict the high pavement temperature at 20 mm depth. A step- wise regression analysis was used to develop the mod- els. The models achieved R2 values of 0.847 and 0.854 with the addition of the air and solar radiation variables, consecutively. The partial F-test indicated that the duration of solar radiation up to the maximum air tem- perature variable for the three variables model should be removed at 0.05 level of significance. 5. The comparison of high pavement temperature model using only air temperature as an independent variable with SHRP and LTPP models indicated that the latter models would estimate higher pavement temperatures 41 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42 20 30 40 50 60 70 80 15 20 25 30 35 40 45 50 Air temperature, oC Pa ve m en t t em p. , o C LTPP SHRP High temp. model Figure 11. Comparison between high pavement temperature prediction model and SHRP and LTPP models for Oman. Therefore, these models would be conserva- tive in the selection of Performance Grade (PG) binders for Oman compared with the developed model. 6. The use of the high pavement temperature model would depend on the availability of data. If solar radiation measurements are available or can be estimated, then the two variable model can be used. Acknowledgments The authors gratefully acknowledge the financial sup- port by Sultan Qaboos University through an internal research grant. The authors acknowledge the help provid- ed by the Project and Maintenance Department in the University for the station installation. References Al-Abdul Wahhab, H.I. and Balghunaim, F.A., 1994, "Asphalt Pavement Temperature Related to Arid Saudi Environment," J. of Materials in Civil Engineering, American Society of Civil Engineers, 6 (1), pp. 1-14. Al-Abdul Wahhab, H.I., Asi, I.M., Al-Dubabe, A. and Ali, M.F., 1997, "Development of Performance-Based Bitumen Specifications for the Gulf Countries," Construction and Building Materials, 11(1), pp. 15-22. Bosscher, P. J., Bahia, H. U., Thomas, S. and Russell, J. S., 1998, "The Relation Between Pavement Temperature and Weather Data: A Wisconsin Field Study to Verify the Superpave Algorithm," Transportation Research Record 1609, TRB, National Research Council, Washington, DC, pp. 1-11. Diefenderfer, B.K., Al-Qadi, I.L., Reubush, S.D. and Freeman, T.E., 2002, "Development and Validation of A Model to Predict Pavement Temperature Profile," Presented at Transportation Research Board 82nd Annual Meeting, Washington DC. Hermansson, A., 2000, "Mathematical Model for Calculation of Pavement Temperatures: Comparisons of Calculated and Measured Temperatures," Transportation Research Record 1699, TRB, National Research Council, Washington, DC, pp. 134-141. Hermansson, A., 2001, "Mathematical Model for Calculation of Pavement Temperatures: Comparisons of Calculated and Measured Temperatures," Transportation Research Record 1764, TRB, National Research Council, Washington, DC, pp. 180-188. Lukanen, E. O., Han, C. and Skok, E. L. Jr., 1998, "Probabilistic Method of Asphalt Binder Selection Based on Pavement Temperature," Transportation Research Record 1609, TRB, National Research Council, Washington, DC, pp. 12-20. Marshall, C., Meier, R.W. and Welsh, M., 2001, "Seasonal Temperature Effects on Flexible Pavements in Tennessee," Presented at Transportation Research Board 80th Annual Meeting, Washington, DC. MINITAB Statistical Software Release 13, 2000, User's Guide, Minitab Inc. Mohseni, A., 1998, "LTPP Seasonal Asphalt Concrete (AC) Pavement Temperature Models," Report FHWA-RD-97-103, Federal Highway Administration, U.S. Department of Transportation, p. 71. Mohseni, A. and Symons, M., 1998a, "Effect of Improved LTPP AC Pavement Temperature Models on Superpave Performance Grades," Presented at Transportation Research Board 77th Annual Meeting, Washington, D.C. Mohseni, A. and Symons, M., 1998b, "Improved AC Pavement Temperature Models from LTPP Seasonal Data," Presented at Transportation Research Board 77th Annual Meeting, Washington, D.C. Neter, J., Wasserman, W. and Kunter, M.H., 1990, "Applied Linear Statistical Models," Third Edition, Richard D. Irwin, Inc., Boston, p. 1181. Shao, L., Park, S.W. and Kim, Y.R., 1997, "A Simplified Procedure for Prediction of Asphalt Pavement Subsurface Temperatures Based on Heat Transfer Theories," Transportation Research Record 1568, TRB, National Research Council, Washington, DC, pp. 114- 123. Solaimanian, M. and Kennedy, T.W., 1993, "Predicting Maximum Pavement Surface Temperature Using Maximum Air Temperature and Hourly Solar Radiation", Transportation Research Record 1417, TRB, National Research Council, Washington, DC, pp. 1-11. 42 The Journal of Engineering Research Vol. 2, No. 1 (2005) 32-42