The objective of this research was to determine if roughness data on railroad crossings could be extracted from INDOT’s road network database in order to determine the need and priority of repair projects. This “railroad roughness” was to be correlated to the general public’s perception of riding comfort by using the panel rating method. From this information a Railroad Crossing Index (RCI) would be developed so that the INDOT could search the database to generate a list of crossings that require repair and prioritize them. To calculate RCI four methods were proposed: 1. The first proposed method for generating RCI equates IRI and RCI. By this method the user would specify the distance over which the RCI (IRI) is calculated, from ten feet to 1/10 mile, centered around the crossing. 2. The second proposed method for generating RCI, was to calculate a difference in IRIs. This was to be done by generating an IRI for a section of road, including the railroad crossing data, then use the same raw data to generate a second IRI after “masking out” the railroad crossing data, and take the difference of the two. 3. The third method proposed for calculating RCI is similar to method one, but does not result in a standard IRI number. In addition to enabling the user to specify the distance over which the RCI is calculated (method 1), the user may also specify the ‘long wave’ parameter used in the IRI calculation. 4. The final method proposed for generating RCI uses a calculated elevation profile. The RCI is the summation of the absolute values of the change in height from a reference index and a moving average of data points surrounding the point under consideration. The repeated failure of the profiler’s computer system resulted in a series of delays and eventually to a complete revision of the computer system and software. During the revision of the software, several changes were made to put the system back in service after a lengthy period of downtime. Unfortunately, some of the changes that were made to the software had an adverse effect on the study. Due to the size of the sampling interval, locating the railroad-highway grade crossings in the data generated by the Department’s profiler was extremely difficult. This problem was compounded by the fact that the Department’s profiler uses a receiver-averaging board to generate a sensor footprint in the height measuring system. Therefore the resolution of the distinguishing features was so poor as to be almost unrecognizable. After locating the railroad-highway crossings in the data, and generating the RCI values for each crossing using the two remaining methods, the results were correlated to public perception of acceptable, or unacceptable, levels of roughness. A logistical regression analysis was performed on the data to see if enough probability of predicting the acceptability/unacceptability of roughness associated with railroad-highway grade crossings existed to validate either of the proposed methodologies for calculating RCI. Both of the remaining methods for calculating RCI failed to meet the minimum amount of probability necessary to validate either method. Also, due to the geometry of the construction of railroad-highway grade crossings, the roughness of the crossing itself is nearly impossible to distinguish from the roughness of the approaches. For this reason, it is unlikely that data collected from any current profiler would be able to provide results sufficient to generate a rating scale based on this technology.

Report Number



roughness, International Roughness Index, IRI, Railroad Crossing Index, RCI, resolution, sample interval, sensor footprint, attenuation, railgap, gauge spacing, rating panel, logistical regression, SPR-2128

SPR Number


Project Number


File Number


Performing Organization

Joint Transportation Research Program

Publisher Place

West Lafayette, IN

Date of this Version