Characterizing Railroad Ballast Using GPR: Recent Experiences in the United States R. Roberts, A. Schutz I. Al-Qadi, E. Tutumluer GSSI University of Illinois at Urbana-Champaign 12 Industrial Way, Salem, NH USA Urbana, IL, USA roger(Zggeophysica1 &om, alan.S(Z4geophysical.com alqadi ui~uc,edu tutumlueguiuc.edu contrasts were readily apparent in the data. More recently, Abstract Recent work has been conducted in the United States starting in 2001, the Federal Railroad Administration has with 2 GHz horn antennas to characterize railroad ballast. There partially funded research to further develop the use of GPR were a number observations made during the course of the for track substructure evaluation [3]. Work initially was project that derive from gaining a more thorough understanding focused on implementation of 1 GHz horn antennas. A of ballast and the interaction of GPR with the ballast matrix. system containing both 400 MHz and 1 GHz antennas has The major observations from over 238 km of track data at four been developed in the UK [4]. Recently, 2 GHz hor antennas different geographical locations include: (1) it cannot be assumed . . ' that there will be a reflection from the bottom of clean ballast or have been used to provide even higher resolution images of that there will be a reflection from the ballast-subballast the ballast [5], [6]. interface; (2) the presence of a strong reflection in the data generally, but not always, infers moderately-fouled to clean There is an obvious progression from the low- to high- ballast above the reflecting boundary; and (3) no observable frequency antennas as the technology evolved. The higher ballast-subballast interface reflections are generally, but not frequency antennas provided increasingly greater resolution always, associated with gradational fouling or a fully-fouled of the ballast structure. Analysis of the high resolution images ballast section. ballast section. from the 2 GHz horn antennas together with data from cross- Index Terms-Ballast Fouling, GPR, 2 GHz Horn Antennas, trenches and 238 km of track from 4 different geographical Void Scattering. locations has revealed interesting correlations and consistencies that relate to anticipated results from lower I. INTRODUCTION frequency antennas. Ground penetrating radar (GPR) is a geophysical tool with a diverse number of applications including but not limited to locating buried tanks, measuring sediment thickness in water Railroad ballast is the uniformly graded coarse aggregate bodies, finding reinforcing in concrete, calculating pavement placed between and immediately underneath railroad ties. thickness and mapping geological strata. It has also been used Subballast, comprised of sand or gravel, provides drainage to characterize railroad track substructure. In contrast to along with the ballast and distributes the applied train loading established applications such as pavement profiling and over the subgrade. The ballast and subballast system is concrete inspection, the railroad application has not become commonly referred to as the granular layer, supporting track routine in the United States. The fact that currently it is not with a design thickness of typically 45-70 cm. In most of the prevalently used in the United States either indicates that it old existing lines, only the ballast is found above the has not been successful in its purpose and/or it has not been subgrade. Rehabilitated and newly constructed lines, economical. Very likely, the lack of enthusiasm with this however, typically include the subballast layer under the technique for mapping track substructure is a combination of ballast. Figure 1 illustrates a typical active railroad cross- (1) moderate success, and (2) an inability to translate the GPR section. There is often a transition area at the ballast- success to economical benefits. Recent developments in GPR subballast interface containing fines. Over time, ballast is technology have permitted unprecedented resolution of the contaminated by aggregate breakdown under traffic by fine ballast layer, which help shed new light on previous GPR materials such as mineral filler and coal dust that fill the void investigations. spaces between ballast stones. This contamination is commonly referred to as fouling. The fouling index is used to A very brief outline of the history of GPR in track describe the level of fouling in track and is calculated as the substructure investigations in the United States provides summation of the percentage of material passing the #4 sieve insight into the current status quo. Perhaps the earliest attempt plus the percentage of material passing the #200 sieve [7]. at using GPR for track inspection was 1985 [1] involving 500 When the fouling reaches a specific threshold, the structural MHz antennas mounted between the rails. The low resolution integrity and drainage ability of the contaminated ballast images produced by the antennas were difficult to interpret, system can be compromised. This leads to track instability, especially in the near-surface. In the mid-1990's, 1 GHz horn which, without corrective maintenance, can lead to train antennas were tested for evaluating railroad substructure [2]. derailments. Hence, early detection of ballast fouling is of These antennas provided enhanced images of the near-surface utmost importance to maintain safe operation. Due to its ballast substructure. Significant electromagnetic property I-4244-0887-3/07/$20.00 © 2007 IEEE 295