New Purdue GPR model could reduce underground pipeline strikes

Researchers at Purdue University have developed a new method to improve the accuracy of ground-penetrating radar (GPR) for locating underground pipelines, a breakthrough that could help reduce utility strikes during construction and excavation activities. 

According to Purdue, the patent-pending technology enhances traditional GPR analysis by applying an uncertainty-aware statistical model that more precisely estimates a pipe’s location, orientation and radius. The work was led by civil engineering professor Hubo Cai and doctoral researcher Yuxi Zhang from Purdue’s Lyles School of Civil and Construction Engineering. By better interpreting electromagnetic wave reflections captured by GPR systems, the model aims to provide contractors and utility-locating professionals with more reliable underground mapping data.

According to the researchers, the system incorporates a Bayesian framework that quantifies uncertainty in pipe detection and creates buffer zones around estimated locations. The approach also includes diagnostic tools to evaluate GPR data quality and improve confidence in detection results, Purdue stated. In testing, the model demonstrated strong predictive accuracy while maintaining lower computational costs.

The research addresses a major challenge facing construction and infrastructure projects. Data from Common Ground Alliance indicates that inaccurate location information accounts for the majority of underground utility damage incidents in the United States, which collectively cause billions of dollars in economic losses each year. The Purdue team says improved detection tools could support safer excavation, better infrastructure mapping and the development of digital models of underground utility networks.

The innovation has been disclosed to Purdue Innovates’ Office of Technology Commercialization, which has filed a patent application for the technology and is seeking industry partners interested in developing or commercializing the system.