Development of ANN prediction model:
The data collected provided a fair insight into the parameters affecting the permeability and voids in PAC mix. From the data collected, relation between voids in total mix and permeability coefficient of PAC was established as in (Fig. 1). Artificial Neural Networks (ANN) were successfully implemented to develop a prediction model from data collected for porous asphalt gradations. The application of ANN in this part was mainly focused on prediction of void percentages in a given PAC mix. A three-layered feed forward error-back propagation ANN architecture was used. From regression plots as given in (Fig. 2), 15-10-1 ANN model was selected as better prediction model to predict void percentages in PAC mix with coefficient of correlation as 0.994 and mean square error of 0.106. Further the test data of 15 samples was provided for evaluation of 15-10-1 ANN model and results obtained in (Fig. 3) depict that the predicted and actual void percentages of PAC gradation test data show good agreement as plotted.
This model was further used to evaluate the effect of the mix parameters on PAC mix to produce different void percentages. In this part, effect of Optimum Binder Content (OBC) was analyzed for PAC mix properties. The results obtained indicates that increase in OBC of PAC mix causes decrease in void percentage for that particular mix.
Analysis of Porous Asphalt Pavement in KENPAVE software:
The result outputs of KENPAVE for each of the cross-sections analyzed gives the idea of variation for maximum deflection, maximum vertical stresses, maximum tensile strain and maximum compressive strain at each point along the depth of pavement. To check for rutting analysis, we consider for Fatigue analysis Maximum Horizontal Tensile Strain (εt) at the bottom of PAC layer and Maximum Vertical Compressive Strain (εz) at the top of sub-grade layer is considered. Then those (εz) and (εt) values helped in estimating maximum allowable number of load repetitions for preventing rutting failure (Nr) and Fatigue failure (Nf) simultaneously for any cross section [16]
It is observed from the graphical analysis that cross-section E is providing maximum allowance for load repetitions in view of fatigue cracking (Nf), that is 2.8E + 06 cycles of single axle dual wheel load. Also, it can be observed that cross-section E is allowing maximum number of load repetitions in case of rutting failure (Nr), that is 2.7 E+08 cycles of single axle dual wheel load and hence the optimum cross-section that is satisfying failure resisting conditions is the cross-section E. So cross section E is considered be the best arrangement of layers in terms of pavement performance. The layer thicknesses of cross-section E are: PAC top layer = 10 cm; Choker course = 5 cm; Stone bed reservoir = 45cm
The result outputs of KENPAVE for cross-section E with different subgrade materials analyzed gives the idea of the damage expected during the service life of pavement along with the damage ratios and expected design life of pavement. The results obtained from the analysis indicate that the design life of pavement with gravelly soil subgrade, which is the most preferred soil for construction of porous asphalt pavement due to its good permeability, is obtained as 15.32 years with least damage ratio of 0.0653. Also, the maximum damage ratio for clay soil is 0.3 and there is 77.74% reduction in design life as compared to the Pavement with gravelly soil subgrade. The allowable number of repetitions for preventing fatigue failure ranges between 7.66 E+06 & 5.31E + 06. The allowable number of repetitions for preventing rutting failure is maximum for gravely soil subgrade i.e., 3.98 E+08 and least for clay soil subgrade i.e. 1.7 E+06
Effect of different axle configurations as well as tire pressures was successfully evaluated for porous asphalt pavement with the help of KENLAYER. Damage analysis was performed for various configurations and tire pressures to evaluate the performance of PAC pavement. The maximum design life of 15.32 years was obtained for single axle dual wheel assembly with contact pressure of 550 kPa. The least design life of 2.37 years was obtained for tandem axle dual wheel configuration with contact pressure of 875 kPa. There is 84.53% reduction in design life of PAC pavement for tandem axle dual wheel configuration with 875 kPa as compared to single axle dual wheel assembly with 550 kPa contact pressure. In case of tandem axle dual wheel configuration, different contact pressures reveal different result. For tandem axle dual wheel configuration, maximum design life is obtained for 480 kPa contact pressure i.e. 9.6 years and least design life of 2.37 years was obtained for 875 kPa contact pressure.
It can be seen from results that maximum damage ratio obtained for bottom of top PAC layer is more for tridem axle configuration and tandem axle assemble with 875 kPa tire pressure. The value of damage ratio at bottom of PAC layer is maximum, i.e. 1.49e − 01, for tandem axle configuration with 875kPa tire contact pressure, followed by tridem axle configuration of 660kPa tire contact pressure with damage ratio equal to 1.03e − 01. In case of damage ratios obtained on top of subgrade, the value obtained is very less than 1. The maximum damage ratio obtained on top of subgrade is equal to 2.03e − 04, indicating least effect of tire pressure on damage of pavement due to rutting.
Analysis of Porous Asphalt Pavement in ABAQUS software:
Prior to modelling of PAP in ABAQUS, comparative study between KENPAVE and ABAQUS was carried out for evaluation of ABAQUS software to model pavement structure and evaluate the critical responses. Results from the comparative study indicate the suitability of ABAQUS software to model PAP with much comparable results obtained in both computer Programme.
The usefulness of ABAQUS was evaluated by modelling of actual pores in PAC layer, which is not otherwise possible in KENPAVE software. Porous Asphalt Pavement was successfully modeled in ABAQUS software with actual pores modeled in top PAC layer. Results obtained from ABAQUS analysis (Fig. 4) show that, with increase in void percentage in PAC mix, vertical stress dissipated on layers under PAC layer decrease. For a void percentage of 16.4%, vertical stress under PAC layer is 267 kPa, whereas for 28% voids vertical stress dissipated below PAC layer is 243 kPa. Consequently, displacement under PAC decreases with increasing percentage of voids in PAC mix.
It is also observed that with increase in percentage of voids in PAC mix, horizontal tensile strain at bottom of PAC layer increases leading to early fatigue failure of pavement. For the lowest void percentage of 16.4%, horizontal tensile strain at bottom of PAC layer is 2.48e-04 and for highest void percentage of 28%, horizontal tensile strain at bottom of PAC layer is 5.53e-04. Similarly number of repetitions for fatigue failure is 7.8e + 05 for 16.4%voids and for 28% voids, no. Of allowable repetition are 5.6e + 04. The impact of horizontal stress is mostly in PAC layer and horizontal tensile stresses under the wheel load increase with increase in the percentage of voids. For void percentages in range of 16 to 19% horizontal strain at bottom of PAC layer is less than 3E-04 and there is not much reduction in allowable number of repetitions to fatigue failure. From (Fig. 5), it is quite evident that void content in PAC layer has more prominent effect on horizontal tensile below PAC layer rather than vertical compressive strains. Since horizontal tensile strain below PAC layer is responsible for fatigue failure in pavement, increasing total void percentage in PAC layer may lead to early deterioration of porous asphalt pavement due to lesser allowable number of repetitions to fatigue failure.