Linear Regression Models to Evaluate Characteristics of Railway-Roadway Level Crossings in a Developing Country

Abstract

Railway-roadway level crossings are dangerous crash prone locations and the safety at these locations is of utmost importance. The objective of this research was to analyze the contributing factors of railway-roadway level crossing crashes. Crash data at railway- roadway level crossings, level crossing characteristics and both railway and roadway characteristics were collected along the coastal line from Pettah to Aluthgama in Sri Lanka. Seventy-seven level crossings were identified along this railway route. The data were collected using field surveys and site visits. The Multiple linear regression analysis was used to compare how a considered set of independent variables affects the dependent variable which was the crash data. According to the multiple linear regression analysis, the distance to nearest curvature from Colombo direction and Galle direction and the available traffic control system at the level crossings were not significant predictors of the occurrence of crashes at level crossings. The sight distances from the seaside and the land-side were significant predictors of crash data according to the regression model results. The sight distance from the seaside had a higher impact than the sight distance from the land side according to the standardized coefficients obtained from the model. The dangerous and hazardous level crossings along the selected route were identified through the detailed analysis of the crash data. Enhancements that could improve the standards of Galvanized Iron (GI) gates were identified and increasing the GI gate operator’s salary was also proposed.