You are on page 1of 18

Back-calculation of pavement layer moduli and forward-calculation of stresses and strains

Authors:

Wei Zhang Technical University of Denmark, Centre for Traffic and Transport
DTU. Building 115, 2800 Lyngby, Denmark Per Ullidtz Technical University of Denmark, Centre for Traffic and Transport DTU. Building 115, 2800 Lyngby, Denmark Abstract: The analytical-empirical (or mechanistic-empirical) method is widely used in pavement design or evaluation. The method has three steps: 1. Determine the moduli of the pavement layers 2. Calculate the critical stresses or strains under the design load(s) 3. Compare critical to permissible values (or use stresses or strains in a deterioration model) If the pavement layer moduli are determined from an inverse analysis (backcalculation) of Falling Weight Deflectometer (FWD) data, then the first two steps in this procedure constitute the analytical part. This paper deals with these two steps in relation to deflections, stresses and strains measured under a FWD. In step 1 and 2 above, the pavement response is calculated using a theoretical model, in terms of deflections in step 1 and of stresses and strains in step 2. All theoretical models involve simplifications with respect to reality and even the most sophisticated theoretical model must be verified against real pavement structures. To verify a theoretical model, the stresses or strains calculated in step 2 may be compared to measured stresses or strains. During the past ten years this has been done on three instrumented pavements in the Danish Road Testing Machine and on three in situ pavements in Sweden. Strains at the bottom of the asphalt layer and stresses and strains at the top of the subgrade (which are frequently used in design methods) were measured and compared to theoretical values. Three different theoretical models were used to first backcalculate the layer moduli and then forward calculate the stresses and strains. The models were: Layered Elastic Theory with linear elastic materials, the Method of Equivalent Thicknesses with a non-linear subgrade and the Finite Element Method where any layer may be non-linear elastic. For the strains at the bottom of the asphalt, reasonably good agreement was mostly found with all theoretical models, although some comparisons gave very large differences. For the strains at the top of the subgrade, the assumption of linear elastic materials gave very poor agreement. A reasonably good agreement could be obtained with a non-linear elastic subgrade. In most cases the best agreement (and the most reasonable values for layer moduli) was found using the Method of Equivalent Thicknesses. 1. INTRODUCTION The analytical-empirical method is widely used in pavement design. The analytical part includes determination of the pavement layers moduli and calculation of the critical stresses or strains under the design load(s). There are several theoretical models available. All theoretical models involve simplifications with respect to reality and even the most sophisticated theoretical model must be verified against real pavement structures.

To verify a theoretical model, the stresses or strains calculated may be compared to measured stresses or strains. During the last decade this has been done on three instrumented pavements in the Danish Road Testing Machine (RTM) and on three in situ test sections in Sweden, (Borrebackevgen). Strains at the bottom of the asphalt layer and stresses and strains at the top of the subgrade (which are frequently used in design methods) were measured and compared to theoretical values. Three different theoretical models were used to first back-calculate the layer moduli and then forward-calculate the stresses and strains. The models were: Layered Elastic Theory (LET) with linear elastic materials, the Method of Equivalent Thicknesses (MET) with a non-linear subgrade and the Finite Element Method (FEM) where any layer may be non-linear elastic. One of the problems encountered in the back-calculation is that the solution from the back-calculation may not be unique, when a pavement has more than two layers (including the subgrade). The output of a back-calculation depends on how the pavement structure is modelled, which program is used, the seed values used, and how the applied load is treated. One way of verifying the validity of a set of layer moduli is to use them in a forward-calculation of stresses and strains and then compare these values to the measured stresses and strains. The test pavements in the RTM and test sections at Borrebackevgen, Sweden, were instrumented with LVDT Soil Deformation Transducers (SDTs) to measure strains in unbound materials, Asphalt Strain Gauges (ASGs) to measure strains at the bottom of the asphalt layer and Soil Pressure Cells (SPCs) to measure stresses in unbound materials. These instruments have been developed over the past 25 years and have been used with success in previous studies [Krarup, J., 1992, Macdonald, R. and S. Baltzer, 1997]. Stresses and strains in the pavement have been measured both under a rolling wheel loading and under FWD loading. This paper focuses on analyzing deflections, strains and stresses measured under FWD loading. Several layered systems are used to model the test pavement in this approach, and programs based on LET, MET and FEM are utilized, both in backcalculation to obtain layer moduli, and in forward-calculation to obtain stresses and strains. Non-linearity of the subgrade is taken into consideration for some models. The calculated stresses and strains are compared to the measured values. 2. TEST PAVEMENTS IN THE DANISH ROAD TESTING MACHINE The Danish Road Testing Machine is a pavement testing facility, with a width of 2.5 m and a length of 27 m. The central 9 m is the actual test section, which is 2 m deep. A plan and section of the RTM are shown in Figure 2.1. The RTM is enclosed in a climate chamber, 4 m wide and 3.8 m in height. Heating and cooling equipment make it possible to maintain a temperature range of -10C to +40C. The ground water level is automatically controlled and may be raised or lowered as desired. The wheel load is hydraulically applied and a single or a dual wheel may be used. The maximum wheel load is 65 kN and the maximum velocity is approximately 25 km/h. 10000 load repetitions, at this load level, may be applied during one 24-hour day. This corresponds to approximately 70000 passages of a standard 80 kN axleload. The lateral position of the wheel can be automatically changed during testing to give a desired transverse wheel load distribution (wander).

Figure 2. 1 The Danish Linear Track Road Testing 2.1 Pavement Structures, Materials and Instruments For the International Pavement Subgrade Study (DRI 1997), the first test pavement (RTM1) was built in 1994, and the second test pavement (RTM2) was constructed in 1997. After RTM2 was subjected to an accelerated loading test with 160,000 load repetitions and two freeze/thaw experiments, a rut depth of 20 mm had developed. After this the pavement was rehabilitated. An open-graded asphalt concrete layer, about 35 mm thick, was placed on the deformed surface. The rehabilitated test pavement was the third test pavement (RTM3). The three test pavement structures are shown in Figure 2.2. RTM1 Asphalt Concrete (AC) Granular Base Course (BC) SubGrade (SG) Drainage Layer (DL) Concrete Bottom (CB) Native Soil (NS) 84 mm 172 mm 1310 mm 180 mm 250 mm AC BC SG DL CB NS RTM2 84 mm 140 mm 1380 mm 180 mm 250 mm AC BC SG DL CB NS RTM3 120 mm 140 mm 1380 mm 180 mm 250 mm

Figure 2.2 Pavement Structures of the three test pavements.

The original asphalt concrete layer was densely graded asphalt. The maximum aggregate size was 16 mm. The AC mix was composed of crushed granite aggregate, sand and filler with a 60 pen. bitumen binder. Mix tests indicated a bitumen content of 5.3 percent and a density of 2420 kg/m3. The base course is a graded crushed natural aggregate with an aggregate particle size ranging from 0-32 mm. It has an optimum moisture content of 6.9 % and a maximum dry density of 2190 kg/m3. The level of compaction varies considerably throughout the pavement between values of 94-97 % of maximum dry density. The relatively low compaction level achieved is due to a low moisture content of around 4 %, and to the fact that the 150 mm granular layer is placed directly on a relatively soft subgrade material. A Danish "Moraine Clay" classified as a clayey silty sand (AASHTO classification A-4(0)) was used as the subgrade. It has an optimum moisture content of 9 % and a maximum dry density of 2045 kg/m3. It is constructed in nine layers of approx. 150 mm thickness each. The reinforced concrete bottom was tested with an FWD before construction of the filter gravel layer. Based on the previous studies carried out in the RTM, the concrete bottom has a modulus of about 27,500 MPa and the drainage layer has a modulus of 250 MPa. For RTM1, the instruments were installed along the centreline of the pavement, while for RTM2 and RTM3 instruments were located in two longitudinal lines, 170 mm on either side of the RTM centreline. The centre-to-centre distance between the dual tires of the loading cart is 340 mm, thus each line of instruments lie along the centreline of one of the dual tires, when the loading cart is aligned along the RTM centreline. The ASGs were installed longitudinally and transversely at the bottom of the asphalt concrete layer. The SDTs for measuring vertical strains were installed in the base course and the three upper subgrade layers. The SPCs for measuring the vertical stresses were placed in the three upper subgrade layers (each layer had a thickness of 150 mm approximately). There were also SDTs and SPCs installed to measure the horizontal strains and stresses in the subgrade, but results from these gauges are not considered in this paper. The vertical distances from the centres of the instruments to the asphalt surface are given in Table 2.1. Table 2.1 The vertical distances from the centres of instruments to the AC surface Instruments Direction or Layer RTM1 RTM2 RTM3 ASG Transverse direction X 84 mm 84 mm 120 mm Longitudinal direction Y 84 mm 84 mm 120 mm (for ex/y) SubGrade Layer 1 (SGL1) 336 mm 304 mm 340 mm SDT SubGrade Layer 2 (SGL2) 476 mm 434 mm 470 mm (for ez) SubGrade Layer 3 (SGL3) 621 mm 574 mm 610 mm SubGrade Layer 1 (SGL1) 276 mm 244 mm 280 mm SPC SubGrade Layer 2 (SGL2) 406 mm 374 mm 410 mm (for sz) SubGrade Layer 3 (SGL3) 546 mm 504 mm 540 mm 2.2 Back-Calculation of the Pavement Layer Moduli Surface deflections of the three test pavements were frequently measured throughout the accelerated loading test. In this study the test results from four measurements are

discussed. The first measurement was conducted on RTM1 after 100,000 load repetitions along the centreline of RTM1 at intervals of 300 mm from position -4.50 m to position +4.50 m. The contact stress was about 500 kPa. The second was carried out on RTM2 after 100,000 load repetitions with a contact stress of 700 kPa, and the last two were conducted on RTM3 after 10,000 load repetitions at 10C and 25C with a contact stress of 500 kPa. For RTM2 and RTM3, the measurements were carried out along the two instrument lines at intervals of 300 mm from position -4.50 m to position +4.50 m. In the back-calculation, dynamic effects of the FWD loading are not taken into consideration. The three theoretical models are Layered Elastic Theory, the Method of Equivalent Thicknesses and the Finite Element Method. Layered Elastic Theory (LET) is the most widely used method for pavement design. The theory of Elasticity is based on the assumptions that the load is static and that all materials are homogeneous, isotropic and linear elastic. Layer moduli of the RTM test pavement have been calculated using the back-calculation program WES (Waterways Experiment Station) [Ullidtz, P., 1998]. Method of Equivalent Thickness (MET) is a simplified version of Layered Elastic Theory. It uses Odemarks method for transforming a layered system into a semiinfinite elastic halfspace on which Boussinesqs equations may be used. Odemarks transformation is based on the assumption that the stresses and strains below a layer depend on the stiffness of that layer only [Ullidtz, P., 1987]. If the thickness, modulus and Poissons ratio of a layer are changed, but the stiffness remains unchanged, the stresses and strains below the layer should also remain (relatively) unchanged. The stiffness of a layer is proportional to: h 3E , 1- m 2 Where h is the thickness of the layer, E is the layer modulus and m is Poissons ratio. The transformation shown in Figure 2.3 should, therefore, not influence the stresses or strains in layer 2, provided that: h 1 E1 1- m 1
2 3 3

h e E2 2 1- m 2

or

E1 1 - m 2 2 he = h1 E 2 1 - m1 2

(1 / 3 )

Where he is known as the equivalent thickness and the suffixes 1 and 2 refer to the first and second layer of the pavement as shown in Figure 2.3.

h1 E1 E2

m1 m2

he E2 E2

m2 m2

Figure 2.3 Odemarks transformation of a layered system.

The transformed system in Figure 2.3 is a semi-infinite halfspace on which Boussinesqs equations may be used, but only for stresses, strains and displacement below the interface. Non-linearity of the subgrade can be taken into consideration with MET. The model used for non-linearity of the subgrade is based on the major principal stress, and has the following format: s1 E = C p
n

(2.1)

Where C and n are material constants, s1 is the major principal stress, and p is a reference stress (atmospheric pressure). Finite Element Method (FEM). The basic concept of the finite element method is that a continuous structure can be treated as an assembly of elements of finite size. The finite element method still assumes the strains to be compatible, i.e. that materials are continuous or solid, except for special elements that may allow for discontinuities. A stiffness matrix is established for each element and a displacement function is selected (usually linear). The coefficients of the stiffness matrices are calculated using the principle of virtual work. The stiffness matrices of the individual elements are combined to form the overall stiffness matrix of the system, which is solved for the known boundary conditions, to give the displacements at all nodes. From this the stresses in the individual elements may be calculated. With FEM any layer may be non-linear elastic. To describe the non-linearity of materials, Equation (2.1) is used as long as the exponent n is negative. For positive values of n, the major principal stress is substituted by the bulk stress. For back-calculation, a 3-layered system is used with MET, and a 5-layered system is used with LET and FEM, as shown in Figure 2.4. 3-layered system with MET AC mAC=0.35 EAC= ? MPa BC mBC=0.35 EBC= ? MPa SG mSG=0.35 ESG= C(s1 /p)n MPa 5-layered system with LET and FEM AC mAC=0.35 EAC= ? MPa BC mBC=0.35 SG mSG=0.35 mSG=0.35 EBC= ? MPa ESG= C(s1 /p)n MPa (FEM) ESG= ? MPa (LET)

CB mCB=0.15 ECB= 27500 MPa NS mNS=0.35 ENS= ? MPa Figure 2.4 The layered systems used in the back-calculation. With all three programs it is usually possible the get a good match between the calculated and the measured deflections. The Root Mean Square (RMS) difference between the calculated and measured deflection basins is usually less than 3 microns. The pavement layer moduli back-calculated from each program are given in Tables 2.2, 2.3, 2.4 and 2.5 for each measurement, respectively.

Table 2.2 The pavement layer moduli back-calculated from the measurement on RTM1 Layer 3-layered system 5-layered system Moduli MET LET FEM EAC MPa 4480 3650 3770 EBC MPa 75 45 55 CSG MPa 14 11 nSG -0.59 -0.78 ESG MPa 21* 61 18* ECB MPa -27500 27500 ENS MPa -2130 234 * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 500 kPa. Table 2.3 The pavement layer moduli back-calculated from the measurement on RTM2 Layer 3-layered system 5-layered system Moduli MET LET FEM EAC MPa 6240 4694 4480 EBC MPa 144 85 108 CSG MPa 32 20 nSG -0.43 -0.63 ESG MPa 38* 93 26* ECB MPa -27500 27500 ENS MPa -1350 250 * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 700 kPa. Table 2.4 The pavement layer moduli back-calculated from the measurement on RTM 3 at 10C Layer 3-layered system 5-layered system Moduli MET LET FEM EAC MPa 10356 10356 9883 EBC MPa 490 232 322 CSG MPa 45 53 nSG -0.37 -0.25 ESG MPa 69* 113 70* ECB MPa -27500 27500 ENS MPa -763 617 * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 500 kPa.

Table 2.5 The pavement layer moduli back-calculated from the measurement on RTM 3 at 25C 3-layered system 5-layered system Layer Moduli MET LET FEM EAC MPa 4208 3936 3766 EBC MPa 203 86 150 CSG MPa 32 24 nSG -0.45 -0.59 ESG MPa 45* 95 36* ECB MPa -27500 27500 ENS MPa -1332 412 * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 500 kPa. It can be seen that a significant difference exists in the pavement layer moduli computed with the three theoretical models. With LET, a low base course modulus and a high subgrade modulus are often obtained. For example, based on the measurement on RTM3 at 25C, the base course modulus computed with MET is twice as high as that calculated with LET, while the subgrade modulus computed with MET is only half of that calculated with LET. To evaluate the validity of the pavement layer moduli, the stresses and strains in the pavement layers were measured under FWD loading. The measurements were performed with a hydraulic plate attached to the underside of the FWD's fixed loading plate, to obtain a more uniform load distribution. The FWD load was moved along the centreline of RTM1 or the two instrumented lines of RTM2 and RTM3 at intervals of 75 mm. The peak response value from each instrument under each drop was registered, and from this an influence line, giving the response as a function of distance, was obtained for each instrument. The stress and strain response was measured under the same condition as the deflections. In the forward-calculation of stresses and strains, the same structures, theoretical models and the pavement layer moduli obtained in the back-calculation were used. Due to the facts that an response (strain or stress) was measured with 2 to 4 gages, and that the pavement layers were not homogeneous, the maximum and minimum measured response are compared to the calculated values in the analysis, rather than the mean values. The number of gages (for each measured response), the measured response and calculated values with the three models are given in Tables 2.6, 2.7, 2.8 and 2.9 for the four measurements, respectively. In the Tables, a negative sign is used for tensile stress or strain; a positive sign is used for compressive stress or strain. Usually, equivalent thickness is multiplied by an adjustment factor, f, to make the calculated values with MET more comparable to values calculated with LET. In this study, the calculated values with MET are obtained with the equivalent thickness adjustment factor included. When the radius of the loaded area, a, is larger than the thickness of the asphalt layer, hac, the equivalent thickness adjustment factor, for the horizontal strain at the bottom of the asphalt layer, will be:

a f = 1 .1 * h AC

0.3

( 2 .2 )

For the vertical strains and stresses in the subgrade the equivalent thickness adjustment factor is 0.8. Table 2.6 The measured and calculated pavement response on RTM1 Strain or Layer or Number of Measured Calculated Stress Direction gages Max Min MET LET ASG X 2 -333 -260 -386 -497 Y 2 -328 -273 -386 -497 (ex/y, me) SGL1 2 2177 1859 2643 923 SDT SGL2 2 1044 778 1169 654 (ez, me) SGL3 2 597 519 664 475 SGL1 2 76 52 56 70 SPC SGL2 2 42 40 37 50 (sz, kPa) SGL3 2 27 27 26 30 Table 2.7 The measured and calculated pavement response on RTM2 Strain or Layer or Number of Measured Calculated Stress Direction gages Max Min MET LET ASG X 4 -304 -242 -282 -404 Y 4 -305 -265 -282 -404 (ex/y, me) SGL1 4 1600 1229 1669 843 SDT SGL2 4 846 595 935 587 (ez, me) SGL3 4 860 800 556 416 SGL1 4 91 60 76 100 SPC SGL2 2 76 63 49 70 (sz, kPa) SGL3 2 39 38 34 50 Table 2.8 The measured and calculated pavement response on RTM3 at 10C Measured Calculated Strain or Layer or Number of Stress Direction gages Max Min MET LET ASG X 4 -100 -63 -69 -114 Y 4 -80 -64 -69 -114 (ex/y, me) SGL1 4 492 264 489 312 SDT SGL2 2 402 213 314 231 (ez, me) SGL3 4 338 191 208 176 SGL1 4 32 29 38 40 SPC SGL2 2 25 20 27 30 (sz, kPa) SGL3 2 22 22 20 20

FEM -440 -440 1665 1000 410 38 30 24

FEM -392 -392 1595 976 403 52 40 31

FEM -106 -106 376 260 135 32 25 19

Table 2.9 The measured and calculated pavement response on RTM3 at 25C Measured Calculated Strain or Layer or Number of Stress Direction gages Max Min MET LET ASG X 4 -158 -131 -145 -227 Y 4 -139 -125 -145 -227 (ex/y, me) SGL1 4 849 554 820 429 SDT SGL2 4 628 386 475 314 (ez, me) SGL3 4 446 207 289 234 SGL1 4 51 44 46 50 SPC SGL2 4 31 27 31 40 (sz, kPa) SGL3 4 28 23 22 30

FEM -204 -204 687 348 173 33 26 20

It can be seen in the Tables above that the calculated vertical strains with LET are much smaller than the measured values. Because the horizontal strain at the bottom of the asphalt layer and the vertical strain and/or stress at the top of the subgrade are of special importance as they are often used as pavement design criteria, a comparison of the strain and stress distribution with the measurement conducted on RTM2 was carried out. Figures 2.5 and 2.6 show the measured and calculated transverse and longitudinal strain at the bottom of the asphalt layer, respectively. Figures 2.7 and 2.8 demonstrate the calculated and measured vertical strain and stress in subgrade layer 1, respectively.
Longitudinal Position mm 50 0 -1500 -1000 -500 -50 -100 -150 -200 -250 -300 -350 -400 -450 0 500 1000 1500

Microstrain

Measured MET LET FEM

Figure 2.5 The measured and calculated transverse strain at the bottom of the asphalt layer of RTM2.

10

Longitudinal Position mm 100 50 0 -1500 -1000 -500 -50 -100 Microstrain -150 -200 -250 -300 -350 -400 -450 0 500 1000 1500

Measured MET LET FEM

Figure 2.6 The measured and calculated longitudinal strain at the bottom of the asphalt layer of RTM2.

2000 Measured MET LET FEM

1500

Microstrain

1000

500

0 -1500 -1000 -500 0 500 1000 1500

-500 Longitudinal Position mm

Figure 2.7 The measured and calculated vertical strain in subgrade layer 1 of RTM2.

11

120 Measured MET LET FEM

100

80

60 kPa

40

20

0 -1500 -1000 -500 0 -20 Longitudinal Position mm 500 1000 1500

Figure 2. 8 The measured and calculated vertical stress in subgrade layer 1 of RTM2. It may be seen that when the equivalent thickness adjustment factor is taken into consideration, the values calculated with MET match the measured values quite well. 3. TEST SECTIONS ON BORREBACKEVGEN, SWEDEN Three test sections instrumented with Asphalt Strain Gauges (ASG), Soil Deformation Transducers (SDT) and Soil Pressure Cells (SPC) were constructed at Borrebackevgen, Malm, Sweden, in 1991. The ASGs were installed at the bottom of the asphalt layer in the driving direction, and the SDTs and SPCs were vertically placed at the subgrade top. Over the last 10 years, 12 series of measurements have been conducted. Pavement deflections were measured with the FWD, and stresses and strains in pavement layers were recorded under a rolling wheel and FWD loading. In this study the results from the measurement carried out in August 1993, are analyzed. 3.1 Pavement Structures and Materials The pavement structures of the three test sections are given in Figure 3.1. Structure III was an open-graded layered system. The structure had a porous asphalt layer on a base course of crushed rock with a grain size of 4-120 mm. The permeability of the structure was very high ensuring that water was quickly drained from the surface even under heavy rain, resulting in a good skid resistance, no splash or spray and a low noise level.

12

Structure I (Traditional Pavement) 135 mm Asphalt Concrete 120 mm Macadam 560 mm Stabilized Sand

Structure II (Asphalt Viacobase) 160 mm Asphalt Concrete (Viacobase) 120 mm Macadam 540 mm Stabilized Sand

Structure III (Open-graded Layers) 190 mm Asphalt Concrete

600 mm Macadam

Geotextile

Native Soil

Native Soil

Native Soil

Figure 3.1 The three instrumented structures at Borrebackevgen, Malm, Sweden. 3.2 Back-Calculation of the Pavement Layer Moduli Surface deflections of the three test pavement were measured at 22C and the contact stress of the FWD load was about 1,000 kPa. In the back-calculation the three theoretical models, MET, LET and FEM, were utilized. The resulting pavement layer moduli are given in Tables 3.1, 3.2 and 3.3, respectively. Table 3.1 Structure I and pavement layer moduli. Asphalt Concrete mAC=0.35 EAC= ? Base Course SubBase SubGrade mBC=0.35 EBC= ? mSB=0.35 ESB= ?

MPa MPa MPa

mSG=0.35 ESG= C(s1/p)n MPa (MET and FEM) mSG=0.35 ESG= ? MPa (LET)

MET LET FEM EAC MPa 3014 3378 2787 EBC MPa 312 194 245 ESB MPa 168 119 138 CSG MPa 55 63 n -0.32 -0.27 ESG MPa 71* 200 89* * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 1,000 kPa.

13

Table 3.2 Structure II and pavement layer moduli. Asphalt Concrete mAC=0.35 EAC= ? MPa Base Course SubBase SubGrade mBC=0.35 EBC= ? mSB=0.35 ESB= ? MPa MPa

mSG=0.35 ESG= C(s1/p)n MPa (MET and FEM) mSG=0.35 ESG= ? MPa (LET)

MET LET FEM EAC MPa 2621 3106 2506 EBC MPa 358 138 229 ESB MPa 141 98 117 CSG MPa 45 47 n -0.41 -0.41 ESG MPa 66* 249 81* * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 1,000 kPa. Table 3.3 Structure III and pavement layer moduli. Asphalt Concrete mAC=0.35 EAC= ? Base Course SubGrade

MPa

mBC=0.35 EBC= ? MPa mSG=0.35 ESG= C(s1/p)n MPa (MET and FEM) MPa (LET) mSG=0.35 ESG= ?

MET LET FEM EAC MPa 1687 2425 2028 EBC MPa 234 127 180 CSG MPa 39 36 n -0.39 -0.40 ESG MPa 70* 195 64* * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 1,000 kPa. With MET and FEM there are only small differences in the base course and subgrade moduli. The base course modulus obtained with LET is rather low, while the subgrade modulus is very high.

14

3.3 Forward-Calculation of the Pavement Response The strains and stresses in the pavement layers were measured under the FWD load. The asphalt layer temperature and the contact stress of the FWD load are given in Table 3.4. Table 3.4 The asphalt layer temperature at the measurement and the contact stress of the FWD load Structure I Structure II Structure III The contact stress of the FWD load (kPa) 530 522 513 Temperature in at surface 26 33 20 at bottom 22 24 19 the asphalt layer C In the forward-calculation, the same theoretical models and the same layered systems were used to compute strains and stresses. The pavement layer moduli obtained from the back-calculation were corrected for temperature using the following relationship [Ullidtz, P. and Sjlin, F., 1994]:

E (T ) = Eref * e

a*(T -Tref )

(3.1)

Where E(T) is the asphalt or macadam modulus in MPa at a temperature of TC, Eref is a reference asphalt or macadam modulus (given in Table 3.5), a is a constant (given in Table 3.5), and Tref is a reference temperature, and Tref =10C. Table 3.5 The reference asphalt or macadam modulus and constant a. Asphalt Concrete Macadam Eref MPa a Eref MPa a Structure I 5678 -0.0808 508 -0.0725 Structure II 5884 -0.0882 387 -0.0589 Structure III 4545 -0.0788 261 -0.0111 Using Equation (3.1) the asphalt layer and macadam layer of the three structures at the temperature of the response measurement are computed and given in Table 3.6.

15

Table 3.6 The temperature used for the calculation and pavement layer moduli after the temperature correction Structure I Structure II Structure III Temperature used for the calculation 24C 28C 19C EAC MPa 2564 1544 2137 EBC MPa 270 251 242 MET ESB MPa 168 141 -ESG MPa 69* 60* 73* EAC MPa 2874 1829 3071 EBC MPa 168 97 131 LET ESB MPa 119 98 -ESG MPa 200 249 195 EAC MPa 2371 1476 2637 EBC MPa 212 161 186 FEM ESB MPa 138 117 -ESG MPa 88* 77* 65* * The subgrade modulus at the top of the subgrade calculated with Equation (2.1) when an applied stress on the AC surface is 1,000 kPa. With MET the equivalent thickness adjustment factor, f, for the horizontal strain at the bottom of the asphalt is determined from Equation (2.2) when hac<a, and f=1 when hac>a. For the vertical strain and stress at the top of the subgrade f=0.8. The measured and calculated pavement response of the three test structures are shown in Tables 3.7, 3.8 and 3.9, respectively. Table 3.7 The measured and calculated pavement response on Structure I Measured Calculated Number of gages Max. Min. MET LET FEM 2 -100 -83 -175 -252 -227 ey microstrain ez s microstrain kPa 2 3 253 31 115 27 239 23 103 21 152 17

Table 3.8 The measured and calculated pavement response on Structure II Measured Calculated Number of gages Max. Min. MET LET FEM 4 -259 -174 -213 -318 -285 ey microstrain 2 365 289 289 86 162 ez microstrain s kPa 3 38 31 25 22 18

16

Table 3.9 The measured and calculated pavement response on Structure III Measured Calculated Number of gages Max. Min. MET LET FEM 4 -111 -72 -141 -163 -147 ey microstrain ez s microstrain kPa 2 3 124 19 102 13 223 18 90 18 152 13

The calculated longitudinal strains at the bottom of the asphalt layers of the three structures are larger than the measured values, for all three methods. The poorest agreement is found with LET, which also produced a low base course modulus. For the vertical strains at the subgrade top, the values computed with MET for structure I and structure II are in the range of the measured values, and for structure III the calculated value is higher than the maximum measured values. While the vertical strains obtained with LET are less than the minimum measured values. 4. CONCLUSION When using Layered Elastic Theory for pavement structures a number of simplifications are made. Pavement materials are normally not solid (or continuous) but consist of particles. Partly because of this, deformation is not only elastic but also plastic, viscous and visco-elastic, and is seldom a linear function of stress. Materials are also inhomogeneous and may be anisotropic. Interface conditions are poorly known and layers are not of infinite horizontal extent. Loads are not uniformly distributed over a circular area and they are not static. Even when some of these simplifications are avoided through the use of Finite Element or Distinct Element methods, there are still simplifications with respect to real pavements, and all methods need to be verified against reality. The attempts at verification described in this paper, show that it is possible to get a reasonable agreement between measured and calculated deflections, stresses and strains, either by using an extremely simple method (the Method of Equivalent Thicknesses, MET, which can be done in a spreadsheet) or through the use of Finite Element Methods (FEM). The critical point in the analysis is that the subgrade must be treated as a non-linear elastic material. If the subgrade is treated as a linear elastic material, which is the case with Layered Elastic Theory (LET), then the vertical strain at the top of the subgrade will be grossly underestimated. This is unfortunate because the vertical strain at the top of the subgrade often is used as an important design parameter. Treating the subgrade as a non-linear material also overcomes the problem of large differences between backcalculated moduli and moduli determined from laboratory testing. When the subgrade is treated as a linear elastic material, the subgrade modulus will often be overestimated by a factor of 2 to 3, because it is determined from the outer geophones, where the stress level is low. When this high modulus is used at the centre of the load (where the stress level is much higher and the modulus therefore much lower) the centre deflection will tend to be too small, and to compensate for this an unrealistically low modulus must be used for the intermediate granular layer. When the subgrade is treated as a non-linear material, much more reasonable moduli are obtained. Although a reasonably good agreement was found in this paper, it was far from perfect. A considerable amount of work remains to be done to improve these methods

17

and this may require that more complex methods be used. It should, however, be kept in mind that increasing complexity is only justified if it leads to real improvement. 5. ACKNOWLEDGEMENTS The authors wish to acknowledge the work of the Danish Road Institute for the cooperation in conducting the research projects in the RTM, and the work of M. Berntman, P. Ekdahl and Malm municipality for the research project carried out in Borrebackevgen, Sweden. 6. REFERENCES Danish Road Institute, Data Analysis Report, International Pavement Subgrade Performance Study, DRI 1997. Krarup, J., 1992, Bearing Capacity and Water, Part I: Materials, Construction and Instrumentation, Note 328, Danish Road Institute, Denmark. Macdonald, R. and S. Baltzer, 1997, Subgrade Performance Study, Part I: Materials, Construction and Instrumentation", Report 85, Danish Road Institute, Denmark. Ullidtz, P. 1987, Pavement Analysis, Developments in Civil Engineering Vol. 19, ELSEVIER, New York, USA. Ullidtz, P., 1998, Modelling Flexible Pavement Response and Performance, Polyteknisk Forlag, Denmark. Ullidtz, P. and Sjlin, F., 1994, Borrebackevgen VI, Syvende mleserie, Institute of Road, Traffic and Town Planning, Technical University of Denmark, Denmark. KEYWORDS Deflections, Strains, Stresses, Moduli, Theoretical models.

18

You might also like