Simulating winter maintenance efforts: A multi-linear regression model

Abstract

Winter Road Maintenance (WRM) ensures road mobility and safety by mitigating adverse weather conditions. Yet, it is costly and environmentally impactful. Balancing these expenses, impacts, and benefits is challenging. Simulating winter maintenance services offers a potential new tool to find this balance. In this paper, we analyze Norway's WRM of state roads during the 2021–2022 winter season and propose an effort model. This model forms the computational core of the simulation, predicting the number of plowing, salting, and plowing-salting operations at any given location over the road network. This is a multi-linear regression model based on the Gaussian/OLS method and comprises three sub-models, one for each of the aforementioned operations. The key explanatory variables are: 1) level of service (LOS), 2) road width, 3) height above mean sea level, 4) Average Annual Daily Traffic (AADT), 5) snowfall duration, 6) snow depth, 7) number of snow days (fallen snow and drifting snow), 8) number of freezing-rain days, 9) number of cold days and 10) number of days with temperature fluctuations. The overall effort prediction accuracy for the winter season 2021–2022 was 71 %. The independent variables, the model's outcomes, and its results when applied to simulate the effects of LOS downgrading on a particular road stretch and estimating CO₂ emission over the whole network, are discussed.