Numerical Modelling of the Hydraulic Performance and Clogging Development of Porous Asphalt Mixes Using CFD and DEM

The hydraulic performance of porous asphalt (PA) pavements is significantly affected by pores connectivity, shape, content, and distribution. The variety of pore patterns has been proven to impact the porous structure significantly. However, for any type of permeable pavement, clogging is a challenge and a potential issue that cannot be perfectly addressed. The accumulated clogging materials, such as debris, sediments, and organic matter, can enter the pores, progressively blocking the water flow path and affecting the infiltration capacity negatively. The functionality of porous asphalt will be eventually lost once the pavement is fully clogged. It is necessary to systematically analyse the relationship between pore characteristics and hydraulic performance, and evaluate the magnitude of the clogging mechanism inside the pore structure. Compared with laboratory test, numerical modelling can provide more details, visualise motions, and reveal the movement at any flow time inside a porous structure, which is hard to observe during laboratory experiments. The discrete element method (DEM) has been widely used to reconstruct the granular structure in the geomechanical field; commonly, soil granules have been represented by spheres. However, there is a substantial difference between spherical particles and actual aggregate in terms of shapes, which might lead to inaccurate structure reconstruction, especially affecting the prediction of hydraulic conductivity. Based on the literature review, limited investigation has been conducted into the hydraulic performance of asphalt pavements considering a more realistic, thus complex, structure of the asphalt concrete. In most cases, the porous structure is created using the DEM approach; however, in those studies, spherical balls representing the aggregate are used to virtually reconstruct a PA pavement specimens. In addition to using simple spheres, previous studies had limitations in reconstructing realistic particle size distributions (PSD). Due to the characteristics of the spherical balls used in past models, compaction of the asphalt sample proved difficult, especially when the size difference between the largest and smallest particle diameters is large. Due to the sphere pattern and PSD, the analysis of virtual PA samples made by DEM using spheres has significant limitations. Two different particle arrangements, cubic and hexagonal arrangements, can also be incorporated in a DEM model. The results from past studies showed that the minimum porosity of spherical aggregates can reach 47.64% for cubic arrangement and 25.95% for the hexagonal arrangement. The design used in this study for the virtual reconstruction of PA specimens follows the guidelines of the Victorian Department of Transport and Planning, Australia (Section 417 - Open Graded Asphalt, 2018) for open-graded asphalt mixes (OGA). According to the specification, OGA's porosity shall be in the range of 18% to 25%, which is outside of the porosity tolerance range for PA samples made by spherical aggregates. To control the air voids content, aggregates in a DEM model would need to be more realistic (i.e. shape and size) so that the compaction process can be similar to the experiments conducted in the laboratory. Page | 1 This thesis aims to investigate the relationship between the hydraulic performance of PA mixes and their porous structure through numerical simulation using the DEM and computational fluid dynamics (CFD). This advanced numerical model combining DEM with CFD method is developed to mitigate the adverse effects brought by spherical shape aggregate and deeply understand the relationship between the volumetric properties of the pore structure of PA pavements and the infiltration capacity. To solve the issue caused by spherical aggregates, a photogrammetry approach is applied to capture the realistic aggregate shape as scanned in the laboratory via a simple mobile phone. The 3D model so reconstructed follows the PSD of the OGA mix, as specified by the Victorian standard. Firstly, the investigation is carried out in 2D to understand the foundations of combining the processes of DEM and CFD . Experimental PA slices were prepared and reconstructed computationally using image morphology for calibration and validation purposes. The result shows that, in a 2D simulation, DEM can perfectly reconstruct the pore structure similarly to the structure captured from the laboratory-made slices. However, it also pointed out that limitations still exist under 2D analysis and a single-phase CFD simulation in pore structure calculations of hydraulic conductivity. To investigate the effects of the volumetric properties and the pore structure of PA pavements on the infiltration performance and run-off potential, a 3D numerical model using DEM for sample generation and CFD for hydraulic simulation is developed. This study scans various aggregates at different gradations to build a comprehensive 3D shape template database. Then, DEM software is applied to randomly generate non spherical particles and compact the asphalt mix to the design porosity of 20% and 25%. To provide a higher-quality and accurate calculation for the water motion inside connected pores, a multi-phase with volume of fluid (VOF) model is applied to replace the single-phase CFD simulation. With the VOF model support, the multi-phase CFD simulation provides detailed information about the water infiltration inside the pore structure dynamically. The effect of the volumetric properties of pore structure on hydraulic conductivity can be further evaluated via the voids distribution curve of pore location over depth. Moreover, the relationship between general air voids and effective voids can also be analysed via VOF-CFD simulation, which builds the fundamental understanding of the effect of air voids on the hydrology behaviour of PA pavements. To investigate the clogging mechanism in pore structure and its impact on the hydraulic performance, the study on the short-term hydraulic conductivity using DEM and CFD approach is conducted. The clogging potential at different porosities is different. Various gradations of sediments are applied to test the clogging potential at 20% and 25% porosity. To better understand the clogging mechanism, pre-tests about clogging simulation with sediment gradations from 0.25 mm to 1.0 mm in PA samples with 20% and 25% porosity are applied. The research shows that it is harder for particles larger than 0.6 mm to travel inside the pores and move deeper into the asphalt layer when the porosity is less than 25% (i.e. surface accumulation). Based on the finding and balance of computing power, the mixed sediments gradation of 80% 0.5 mm and 20% 0.75 mm Page | 2 particles is used for the analysis of the clogging mechanism and reduction in hydraulic conductivity. Lastly, it should be noted that the study of 2D modelling with single-phase simulation (Chapter 4) is published in the Journal of Hydrology in April 2023. The study of 3D modelling with multi-phase CFD simulations (Chapter 5) has been submitted to Construction and Building Materials in October 2023, and it is currently under review. The study “Investigation for the effect of sediment clogging development on the short term hydraulic conductivity using CFD and DEM method” (Chapter 6) prepared in December 2023 is currently under review.