Validation of Source Approval of HMA Surface Mix Aggregate Using Spectrometer: Final Report

016121482016Final ReportPDFTech ReportMD-16-SHA/MSU/4-2SHA/MSU/4-2AggregatesElectromagnetic spectrumHot mix paving mixturesNeural networksQuarriesHighwaysMaterialsPavementsHMAaggregatespectroradiometerchemometricsneural networkResearch HubHighwaysMaterialsPavementsUnited StatesMaryland. State Highway Administration. Office of Policy & ResearchWilson, Frederick K.Owolabi, OludareWilloughby, ArthurWhitney, James IIMorgan State UniversityMaryland. State Highway Administration. Office of Policy & ResearchMid-Atlantic Universities Transportation CenterUniversity Transportation Centers Program (U.S.)US Transportation CollectionThe main focus of this research project was to develop methodologies for the validation of source approval of hot mix asphalt surface mix aggregate. In order to further enhance the validation process, a secondary focus was also to create a spectral library. Two methodologies were developed as part of this project: Chemometrics, using GRAM/IQ software and Neural Network (NN). Spectra of aggregate samples from 19 quarries were extracted, using the portable ASD FieldSpec 4 spectroradiometer (FS4). Because the aggregate samples when placed in a petri dish form heterogeneous surfaces, it was necessary to utilize specialized equipment (ASD FS4 Turntable) which converts the surface of the aggregate samples in the petri dish to a homogeneous surface and enhances the spectra extraction process. Ten spectra were extracted from each aggregate sample and then averaged to produce one spectrum. In all, three spectra were extracted per aggregate sample, which ensures representativeness of the aggregate samples. All the spectra were collected in reflectance unit and the data was exported into ASCII for further analyses. Results for both methodologies were encouraging. The Chemometrics was also able to discriminate limestone based on the quarry locations. It also correctly classified 100% of aggregate samples that were not from the aggregate sample population and explained the frictional and physical variability of aggregates within a given quarry over a period of time, making it useful as a diagnostic tool to validate aggregate source. The NN, which relies heavily on the variance of the training set, was able to provide outstanding parameter estimations. Thus the NN is a viable solution for providing quick parameter estimations, based solely on the optical spectrographic measurement of the aggregate sample