This project developed a methodology to extract layer elastic modulus/stiffness from composite soil stiffness and GPS-based position provided by currently available vibratory IC rollers. The developed methodology combines two key components that were advanced in this investigation, namely, forward modeling and inverse analysis. Forward modeling efforts focused on finite element (FE) and boundary element method (BEM) techniques to predict roller-measured composite stiffness values for ranges of layer elastic moduli and layer thickness expected in practice. Inverse analysis or back-calculation works in reverse and provides an estimate of individual layer elastic modulus using IC data. The back-calculation effort in this investigation focused on developing a real-time approach to modulus prediction and characterizing the uncertainty in estimated layer moduli based on measurement error and parameter sensitivity. Both FE and BEM forward modeling approaches were successful in explaining the relative influence of layer properties (layer modulus and layer thickness) on roller-measured composite soil stiffness. Forward model results matched relatively well with experimental data. Inverse analysis was pursued with a traditional gradient search approach that proved successful but time-intensive when using the FE or BEM forward models. As a result, inverse analysis using FE or BEM forward models cannot provide real-time back-calculation of layer moduli using current computing power. The development and use of a statistical regression forward model proved successful in capturing the essence of the FE and BEM results and in enabling real-time inverse analysis to estimate layer modulus values. The investigation demonstrated that layered elastic modulus can be estimated from IC data over a wide range of layered earthwork configurations (layer thickness and ratio of layer moduli). The methodology can be implemented via software algorithms that can be integrated into any commercially available IC software offered by roller manufacturers, consultants and third-party vendors, e.g., navigation system providers. IC software is used on-board the roller and/or on mobile and desktop computers. Therefore, the implementation of the methodology would be performed by the IC software companies. Alternatively, software algorithms could be employed independent of existing IC software. In this approach, the IC data (composite soil stiffness and GPS coordinates) from commercially-available software would be fed into a separate program that would provide layer moduli. The implementation of this latter approach could be performed by any interested party. The methodology generated is generic and can be applied to any currently-available proprietary measures of ground stiffness from vibratory rollers.
This contractor's final report is available.