PAVDAM - probabilistic and volumetric design of asphalt mixtures

[摘要] ENGLISH ABSTRACT:The dissertation presents a design procedure for HMA mixes based on probabilisticand volumetric approaches, hereafter referred to as PAVDAM. Central to PAVDAMis the use of an analytical model for estimating the voids in the mineral aggregate(VMA) of asphalt mixes. Validation of the mix design procedure was done through acceleratedpavement testing (APT) with the Model Mobile Load Simulator (MMLS3)at the National Center for Asphalt Technology (NCAT) test track in Opelika, Alabama.In addition, the semi-circular bending (SCB) test was evaluated to serve asan analysis tool to augment the proposed mix design procedure. Non-linear finite elementanalyses using a quasi-isotropic material model were done using the CAPA-3Dfinite element system developed at the Technical University of Delft in the Netherlandsto better characterise the tensile strength properties of specimens tested withthe SCB.PAVDAM is partly based on performance related and analytical procedures suchas the Stategic Highway Research Program (SHRP) Superpave and the Belgium RoadResearch Centre (BRRC) design method. The mix design system was developed basedon gyratory compaction procedures. In this regard, the criteria relating gyratorycompaction to design traffic as specified in Superpave are used. PAVDAM differsfrom other mix design methods in that a probabilistic approach is used to accountfor the variabilities associated with mixture components and properties during themanufacturing stage. It serves as a subset of the volumetric optimisation stage of themix design process.The development of an analytical model to estimate the VMA of an asphalt mixwas the central theme of the dissertation. The analytical model developed is basedon modified Aim and Toufar (MAT) packing models developed under SHRP researchand used in the concrete industry. The MAT packing models are based on the theoryunderlying the packing of monosized spheres and the combination of successive binarysystems. These models were further refined for use in the asphalt industry. A probabilisticprocedure based on the BRRC PRADO packing model is used to account forthe influence of size ratio of the successive monosized binary systems. The result wasa model that allows an estimation of the VMA of a mix from the gradation of themix, the voids in the filler and the porosities of the individual aggregate fractionsmaking up the mix. Research was undertaken to couple VMA estimates with gyratorycompaction levels. This allowed estimates to be made of the optimum bindercontents of mixes for different design traffic levels.The VMA of a mix is difficult to estimate accurately since it is difficult to quantifythe factors influencing VMA such as gradation, particle shape, angularity, texture andrugosity. Furthermore, the influence of binder content and compaction must be takeninto account. The MAT packing model underestimates the VMA of mixes comparedto measured values. For this reason it is necessary to calibrate the model to allowmore accurate estimations. More sophisticated models are required to more accuratelyestimate the VMA of mixes. It is recommended that the development of these beexplored further.Asphaltic materials are inherently heterogeneous and there are a large number offactors that influence their volumetric properties. Because of this, Monte Carlo simulationtechniques are used in PAVDAM to evaluate the combined effect of the variabilitiesof significant material properties. The dissertation expands on the differentvariabilities and the effects of variability on mixture volumetrics and mix design verification.The dissertation outlines the algorithms and procedures used in PAVDAMto estimate the binder content of a mix. In order to validate the PAVDAM model,analyses were done to determine the reliability of specific NCAT MMLS3 test sectionmixes in terms of densification in the field. A comparison of PAVDAM estimatedand field binder contents allowed a ranking of the reliabilities of the different sectionmixes in terms of field densification at the design traffic level. This ranking comparedfavourably with that obtained from an analysis of actual densification trendsmonitored in the field under full-scale trafficking.Initial FEM analyses of the SCB using linear elastic isotropic modelling allowedthe development of equations to characterise the tensile strength and modulus characteristicsof specimens tested using the SCB. It was emphasized that these equationsdo not provide a realistic assessment of the strengths or moduli of asphaltic materials.The strengths and moduli of these materials are influenced by strain rates within thematerials that cannot be assessed using a simple linear elastic approach. To addressthis, an alternative FEM analyses using CAPA-3D was undertaken. An approach wasadopted to account for the influence of tensile and compressive strain rates on modulus.The analyses made use of a quasi-isotropic material model and it was shown tobetter characterise the tensile strengths of HMA materials using the SCB. The analysesalso indicated that the tensile strengths determined using the equations initiallydeveloped based on a linear elastic approach result in strengths that are unrealisticallyhigh. It is recommended that further finite element research be done using non-linearmaterial modelling to evaluate the very complex stress-strain conditions within anSCB specimen to better characterise fracture response. It is also recommended thatthe fatigue characterisation of HMA be explored based on strength tests using theSCB.PAVDAM represents a rational approach to mix design, a shift from experimentalempiricism towards scientific fundamentalism. PAVDAM can be used to define thespatial composition of asphalt mixes. The influence of mix component variability maybe addressed and reliability assessments of candidate gradations are possible duringvolumetric optimisation. Furthermore, changes in the volumetric properties of asphaltmixes may be investigated. As such, PAVDAM is a mix design management tool andcan only be effective when used as part of a system that closely monitors variabilityand systematically refines the underlying packing model.

[发布机构] Stellenbosch University