The measurement of density or void ratio during the compaction of geomaterials (soils and unbound granular materials) in the field during road construction is essential for superior performance. The specifications adopted by the road authorities worldwide are exclusively based on density. However, estimating density evolution proximally or non-destructively is challenging. Conventional field-based density measurement techniques are hazardous, slow to use and are point-based measurements.
This study developed a novel methodology to estimate the density of geomaterials non-destructively in real-time during the compaction process. The methodology included measuring the surface deformation using Light Detection and Ranging (LiDAR) systems attached to rollers and developing physics-based 1-Dimensional and machine learning (ML) based constitutive models to relate the measured parameters to the density. The developed methodology was validated in an indoor environment where a large soil box (dimensions: 7.5 m×4 m×0.8 m) was fabricated and a well-graded sand in 5 layers of 100 mm was compacted using a 1.5-tonne instrumented roller. The measurement of deformation provided an opportunity to estimate the density in real-time. The estimated density using 1-D model and a ML based classification model had an error of 20% and 16% respectively when compared to density measured from Nuclear Density Gauge (NDG).
This novel instrumentation allowed the density to be measured during compaction with high accuracy, which presents an unprecedented advantage over other conventional approaches, which are intrusive and pointwise, thereby ensuring that the road will be constructed expediently and will function satisfactorily, minimising the occurrence of premature failures. The continual measurement of density during compaction will also facilitate maintaining uniformity of the density, thereby reducing the potential for excessive differential deformations.